This week on Eximius Echo, we’re looking at how founders can build a smarter AI cost stack without overpaying for capability they do not need.

The end user sees one answer. Your product may have paid for five things to produce it. A support query can trigger retrieval, classification, reasoning, a tool call, verification and then the final response. An agent may loop through several of those steps before anything even appears on the screen.

That is why AI cost is becoming a product-design decision. For founders, the useful question is no longer “Which model is best?” It is: what level of intelligence is this workload worth paying for?

INTELLIGENCE IS BECOMING A BUDGETING DECISION

Model choice has expanded in two directions at once. Frontier models keep getting more capable, while cheaper and faster models are becoming good enough for a growing set of production tasks.

The gap matters because capability, speed and cost do not move together. Look at how differently that trade-off plays out across some of today’s leading models:

The point is not to pick a winner. The point is that founders can now buy very different combinations of intelligence, speed and cost. Paying for the strongest model everywhere is a choice, not a default.

That creates a new operating discipline: build an intelligence budget.

For an AI-native startup, this budget can sit directly inside gross margin. A model decision that feels trivial at 1,000 requests can become material at 100,000, especially when each request contains several hidden calls.

SO HOW SHOULD FOUNDERS BUILD THAT INTELLIGENCE BUDGET?

A useful intelligence budget comes down to four decisions for every AI workflow:

1. Set The Quality Floor

Start with the outcome, not the model.

If you are extracting fields from invoices, classifying support tickets or summarising routine documents, “good enough” may sit well below frontier intelligence.

If you are generating production code, making a financial recommendation or giving an agent permission to act, the cost of being wrong rises sharply.

The model only needs to be as capable as the consequence demands.

2. Move Beyond Cost Per Token

Model API pricing is easy to compare. It is also easy to over-index on.

Two models can charge very differently per token and still produce very different economics once the task is completed. One may use more reasoning tokens. Another may need retries. A cheaper first attempt can become expensive if it needs repeated correction.

So move the unit of economics forward:

cost per token → cost per task → cost per successful outcome.

That last number is what ultimately hits the business.

3. Route Intelligence, Do Not Standardise It

A startup does not need one model to do everything.

Routine requests can go to a cheaper default. Difficult cases can escalate. Repeated context can be cached, prompts can be trimmed, and reasoning depth can increase only when the workflow demands it.

And not every step needs generative AI in the first place. Deterministic tasks can often be handled more cheaply through traditional ML or rule-based engines, reserving LLMs for the parts of the workflow that need reasoning or generation.

This becomes especially important for agents. One visible user action can translate into planning, retrieval, tool use, retries and verification underneath. Every unnecessary loop adds tokens and cost.

The architecture behind the interface increasingly determines the bill.

4. Know What You Are Optimising For

Cost is one variable in the budget.

A consumer product may trade higher model cost for lower latency. A document-heavy workflow may pay for longer context. A regulated enterprise may accept higher infrastructure cost for privacy or self-hosting. A coding product may spend more on reasoning if it materially reduces failed runs.

The cheapest architecture is not always the most economical one. Know which constraint your product is paying to solve, and whether that premium is worth it.

If you are paying frontier-model prices for work that a cheaper model can reliably handle, or spending tokens your workflow does not need, then yes: you are overpaying for AI.

THE BOTTOM LINE

The model race is creating abundance. For founders, that creates an allocation problem.

The advantage will come from knowing where premium intelligence materially improves the outcome, where cheaper intelligence clears the bar, and how to move between the two as the product scales.

The winning AI product may not use the smartest model everywhere. It will know exactly where intelligence is worth paying for.

If you’re building the next generation of AI products, we’d love to hear from you. Write to us at pitches@eximiusvc.com.

This week on Eximius Echo, we’re looking at what happens when AI starts planning travel around the traveller.

You tell an AI you want five days away with your partner, somewhere warm, under ₹2 lakh, with great food, minimal transit, and enough time actually to relax.

It already knows you prefer boutique hotels to resorts, hate red-eye flights, avoid packed itineraries, and spent your last holiday wishing you had booked fewer activities.

So it does the work: narrows the destination, builds the itinerary, compares stays, factors in crowds and travel time, books what you approve, and replans when it rains.

That is where travel is heading.

Today, travel remains one of the internet’s most fragmented consumer journeys. Discovery happens across Instagram, Google, YouTube and recommendations from friends and family. Flights, hotels and activities sit across different platforms. Planning means comparing dozens of variables, while support often disappears once the booking is complete.

AI changes the interface from search to intent. In that world, the advantage shifts from having the largest catalogue to making the best decision for a specific traveller.

HOW THE MARKET IS RESPONDING

The first model is emerging from incumbents that already own distribution, inventory and transaction volume.

Booking.com has expanded AI across trip planning, natural-language property search, review summaries and property Q&A. Expedia is pushing further across planning, comparison, booking, service and in-trip assistance, including tools that compare properties around variables like vibe, location and trade-offs.

In India, MakeMyTrip’s Myra takes users from conversational discovery to transaction-ready options across flights, hotels and ancillary services. It handles over 50,000 conversations daily across eight languages, with more than 45% of queries coming from Tier-2 and smaller cities. Ixigo, meanwhile, now offers an AI trip planner alongside its existing flight, train, bus and hotel marketplace.

The incumbent playbook is taking shape: use AI to make existing supply easier to discover, compare and transact.

Newer players can start somewhere else.

Rather than adding AI to a large catalogue, AI-native and niche platforms can build around a specific traveller, trip type or workflow from day one. That opens several models:

THE REAL UNLOCK IS CONTEXT

Adoption is already visible: 33% of Indians use AI for trip planning.

But generating a neat itinerary is only the starting point.

A product that remembers who you travel with, your preferred pace, budget, food choices, hotel style, loyalty memberships and previous trips can make increasingly better decisions each time.

That opportunity becomes especially interesting in India. While international trips account for far higher spend, 99% of trips taken by Indians remain domestic. At the same time, 92% of Indian travellers belong to at least one loyalty programme, with 37% using those benefits for leisure travel.

Trust remains the constraint. Holidays are expensive, infrequent and emotionally loaded. A bad recommendation can cost money, time and a meaningful chunk of annual leave.

ONE MODEL WE’RE SEEING UP CLOSE

30 Sundays, which we first backed at pre-seed, is taking the AI-native concierge route.

Its philosophy is simple: “Learn First, Book Second.”

Customers get personalised, video-first itineraries that surface crowds, queues, travel time, food options and the actual pace of a day. Travellers can modify itineraries through voice, while an AI-plus-human model supports planning, booking and on-trip assistance. AI also powers lead qualification, itinerary creation, follow-ups and reservations.

Monthly Gross Booking Value (GBV) has grown from ₹1 Cr in May 2025 to ~₹16 Cr by June 2026, and the company recently raised ₹61 Cr in a Series A led by Bessemer Venture Partners, with continued participation from Info Edge Ventures and Eximius.

THE BOTTOM LINE

The market is converging on the same idea from different directions.

Incumbents are making massive travel marketplaces more intelligent. AI-native players are rebuilding narrower journeys around context from the ground up.

The next battle in travel will come down to who understands the traveller best, and who can turn that understanding into action.

If you’re building at the intersection of Consumer AI, travel, or agentic experiences, we’d love to hear from you. Write to us at pitches@eximiusvc.com.

This week on Eximius Echo, we’re exploring why personal computers are getting smarter and what that means for you.

For over fifty years, personal computers have relied on a familiar division of labour. The Central Processing Unit (CPU) handled instructions in sequence. The Graphics Processing Unit (GPU) processed thousands of calculations in parallel. Together, they powered everything from spreadsheets to video games.

Now, they’re being asked to do something entirely different: run AI continuously, privately, and in real time.

How We Got Here

In 1971, Intel’s 4004 began the microprocessor era. Over the decades that followed, CPUs grew faster, NVIDIA’s GeForce 256 introduced the modern GPU, CUDA (Compute Unified Device Architecture) turned graphics processors into the parallel computing backbone of AI, and Apple’s M1 showed what became possible when the CPU, GPU, and memory were designed as one system. By 2024, Microsoft required Copilot+ PCs to deliver at least 40 TOPS, or trillion operations per second, of neural processing power.

The personal computer had become an AI device. Its architecture was still catching up.

Enter Vera Rubin

NVIDIA’s Vera Rubin is a tightly coordinated AI platform that combines purpose-built CPUs, GPUs, networking, and storage into one system. A single rack connects 72 Rubin GPUs with 36 Vera CPUs and is designed to run AI inference at up to one-tenth the cost of the previous generation.

Vera Rubin itself belongs in data centres, not laptop bags. Its significance lies in the design principle behind it: AI performance depends on how efficiently the entire system moves data, shares memory, and coordinates workloads, not simply how fast an individual chip can calculate.

That principle is now reaching personal computers through NVIDIA’s RTX Spark platform. With up to 128GB of unified memory and one petaflop of AI performance, RTX Spark can run larger models and more demanding AI workflows directly on the device.

The result is a PC designed around personal AI agents rather than occasional AI features.

What This Means for You

1. A private assistant on your laptop

Local AI models can work across your documents, emails, notes, and meetings without sending sensitive information to a remote server. Whether you’re reviewing contracts, organising research, or summarising personal files, your data stays on your machine.

What this unlocks: Privacy becomes a hardware feature, not just a software promise.

2. AI that works without the internet

Writing support, document search, translation, and file analysis can continue on a flight, in remote locations, or wherever connectivity is unreliable. Your AI no longer depends on reaching a cloud server before it can help.

What this unlocks: Your assistant works wherever your laptop does.

3. Responses that feel immediate

When the model and its data live on the same device, every interaction skips the network round trip. That reduces latency and makes AI feel less like a service you’re calling and more like a built-in capability of the computer itself.

What this unlocks: AI starts behaving like an integrated computer function rather than a service you wait for.

4. Creative tools that understand the whole project

More memory and local AI compute allow laptops to run larger AI models than they could before. That gives creative tools the ability to work across larger videos, image libraries, codebases, and research archives without relying on the cloud for every task.

What this unlocks: Creative tools that understand the entire project, not just the latest prompt.

5. Small language models for personal and company use

More powerful personal computers could allow companies and individuals to design and run Small Language Models (SLMs) tailored to their own data, knowledge, and workflows. This could be especially valuable in regulated industries, where privacy, security, and compliance requirements can limit the use of general-purpose cloud-based AI.

What this unlocks: Purpose-built AI designed around the specific needs of an individual or organisation.

The Bottom Line

Personal computing has always evolved in the same rhythm: hardware gets better first, software catches up next, and consumers only realise the shift once entirely new experiences become possible.

We are at the beginning of that cycle again.

For years, cloud AI has been the default because personal computers were not built to run capable models on their own. That constraint is beginning to disappear. As more intelligence moves onto the device, AI becomes faster, more private, and available wherever your computer is.

If you’re building the future of technology, we’d love to hear from you. Write to us at pitches@eximiusvc.com.

This week on Eximius Echo, we’re looking at how the AI infrastructure stack is being reworked to get more performance from every GPU.

For the last few years, AI infrastructure has been built around training: larger models, bigger clusters, more GPUs. This defined the first wave of the AI buildout.

The next wave is being shaped by inference: running models continuously across products, workflows, enterprises, and software systems. This year, inference is expected to account for roughly two-thirds of AI compute, up from one-third in 2023 and half in 2025.

Agents are accelerating this shift. A chatbot answers a prompt. An agent plans, retrieves context, calls tools, switches between models, checks outputs, and repeats until the task is complete. Each step adds inference demand. Each workflow adds latency pressure.

Efficiency gains are not reducing demand either. Quantisation, distillation, and mixture-of-experts architectures can cut compute cost per inference token by 40% to 80%. However, lower costs make more use cases viable, expanding the total surface area for AI.

This creates the infrastructure paradox of the agentic era: GPUs are more powerful than ever, yet supply constraints, data movement, and scheduling inefficiencies still leave significant capacity underused.

Three answers are emerging, each focused on a different layer of the stack.

Approach #1: RETHINK THE HARDWARE ITSELF

Today’s GPUs were built for more predictable AI workloads: train a model, run batches of inference, move data through the system. Agentic AI behaves differently. It keeps context alive, moves between models, runs longer sessions, and demands lower latency at every step.

That creates a hardware problem. Each time a system reloads model weights or moves data between memory and compute, performance leaks, latency rises, and energy gets wasted.

This approach asks a sharper question: what would inference hardware look like if it were designed specifically for agentic workloads?

The answer is purpose-built silicon where multiple models can stay resident on-chip, session memory can persist separately from compute, and compliance checks can happen closer to the hardware layer. The aim is to reduce the cost of switching between models and make long-running, multi-model AI systems faster and more energy efficient.

Company to watch: Phynomy, which is building inference chips for agentic AI with multi-model residency and hardware-level compliance validation.

Approach #2: FIX THE DATA PIPELINE

The argument here is simple: the chip may be fast enough, but the system feeding it is not.

Modern inference workloads depend on datasets, vector stores, file systems, retrieval layers, and memory hierarchies. When data arrives late, GPUs wait, and that’s expensive.

As inference becomes more data-intensive, utilisation becomes a data movement problem. The bottleneck is no longer only how fast the GPU can compute, but how reliably the right data reaches it at the right time.

The focus is the path between storage and compute. Intelligent caching, AI-native storage, and workflow-aware orchestration keep data staged before the GPU needs it. The goal is to reduce idle time and get more output from the same infrastructure without changing the chip.

Company to watch: TensorMem, which is building an AI-native data acceleration layer aimed at closing the “AI Memory Wall.”

Approach #3: OPTIMISE HOW WORK GETS SCHEDULED

Modern AI applications rarely run one model at a time. A single agentic workflow may call a speech model, an LLM, an embedding model, a reranker, a vision model, and a retrieval system in sequence.

Today, these models are often deployed as separate services. Each service may hold dedicated GPU capacity that sits idle between tasks. The result is fragmented infrastructure, even when total demand is high.

This camp treats inference like an operating system problem. Instead of reserving fixed GPU capacity for individual models, compute is allocated dynamically across workloads in real time. GPUs become a shared pool that can move with demand.

The goal is to raise utilisation, reduce inference costs, and make multi-model workflows easier to run at scale.

Company to watch: Neurafewz, which is building an inference OS with a compiler and runtime scheduler that shares GPU resources across multi-model workflows.

Why These Layers Reinforce Each Other

These approaches operate at different layers of the stack.

A future deployment could look like this: an inference OS schedules multi-model workloads, a data acceleration layer keeps inputs ready before the GPU needs them, and next-generation silicon executes the workload with lower latency and less energy waste.

Each layer compounds the value of the others. Scheduling improves utilisation. Data acceleration removes upstream stalls. Purpose-built hardware reduces the cost of execution.

The companies building at each layer may sell to different budgets today, but the strongest AI infrastructure stacks will bring these pieces together. As inference becomes the dominant AI workload, the winners will be the teams rebuilding every layer that determines how much value each GPU can deliver.

If you’re building at the intersection of AI infrastructure, inference optimisation, compute efficiency, or agentic systems, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

This week on Eximius Echo, we’re looking at a shift sitting at the intersection of all three: Agentic Payments.

For decades, the evolution of commerce has been a story of reducing friction. Cash gave way to cards. The internet birthed e-commerce. Smartphones put entire stores into our pockets. UPI made payments instant and habitual.

But through all these eras, one bottleneck remained constant: a human still had to click “buy.”

Today, we are entering the era of Agentic Commerce and Agentic Payments, where AI transitions from a passive assistant to an autonomous economic agent with its own wallet, permissions, and spending power.

This is the dawn of the “Do It For Me” economy.

The Engine Room: How Agentic Infrastructure Works

The core shift in agentic payments is moving from a “human-in-the-loop” model, where a person approves every step, to a “human-on-the-loop” model, where humans set the guardrails, budgets, and policies while AI handles routine decisions.

The infrastructure is evolving across four models:

1. Human-Assisted Checkout: The agent finds what you need, but stops at checkout for manual card entry.

2. Stored Credentials: The agent uses tokenised cards “on file” to pre-populate checkout pages, requiring a final tap of approval.

3. Virtual Cards: This is where autonomy begins. The agent is issued a single-use or budget-capped card with programmatic controls that allow it to pay within strict limits.

4. Stablecoin Wallets: The agent uses programmable wallets and stablecoins to conduct instant, borderless, machine-to-machine payments without traditional checkout flows.

Each model gives the agent more autonomy. The real unlock begins when payment authority becomes programmable, revocable, and auditable.

B2B Agentic Payments: Shifting from Hours to Outcomes

Traditionally, B2B payments operate like a slow relay race of initiation, approval, authentication, settlement, reconciliation, and reporting. Agentic AI collapses this into a closed loop in which payments are determined by context, policy, and real-time business conditions.

Some of the strongest use cases, with immediate ROI, are already clear:

The B2B bottleneck is data quality. If an agent operates on fragmented or dirty data, it will make poor financial decisions faster. Before enterprises automate payment execution, they need cleaner data architecture, stronger controls, and audit-ready workflows.

B2C Agentic Payments: The Rise of the AI Concierge

In the consumer space, agentic payments promise to reduce the friction of choice overload.

You set the parameters: a direct flight under a certain budget, an insurance renewal only if the premium drops, or a product purchase only during a sale. The agent handles discovery, comparison, timing, and payment.

Two use cases stand out:

But B2C has the harder trust problem. Merchants will need to decide whether to welcome AI shoppers, especially if agent-led purchases create higher returns, disputes, or chargebacks when consumers say, “I didn’t buy that, my agent did.”

Getting Platforms Agent-Ready

If an AI agent cannot see your products, read your terms, or verify your checkout, your business risks becoming invisible in the agentic economy.

Companies will need to prepare across four layers:

1. Expose Verifiable Surfaces: Catalogues, pricing, availability, return policies, and negotiation terms must become structured and machine-readable.

2. Audit the Handoffs: B2B companies need to identify human-dependent steps across procurement, payments, reconciliation, and approvals before handing agents payment authority.

3. Master Agent Marketing: Traditional marketing is built for human psychology. AI agents care about price, availability, trust signals, and policy fit. Brands will need to optimise for agents too.

4. Build the Trust Stack: Agent identity, scoped permissions, spend limits, anomaly triggers, and dispute-ready audit trails will become core infrastructure.

The Dark Side: Risks and the New Liability Fog

Giving software a wallet introduces a new class of risk, as follows:

1. Liability: If an agent buys the wrong inventory, overpays a vendor, or renews an unwanted subscription, who is responsible: the user, the developer, the merchant, or the payment network?

2. AI-Driven Fraud: Agentic AI could scale synthetic identity fraud, refund abuse, social engineering, and chargeback manipulation.

3. Frictionless Debt: If AI spends continuously in the background, the psychological “pain of paying” weakens. Consumers may lose visibility into subscriptions, renewals, and recurring obligations.

The industry will need new legal frameworks, agentic chargeback codes, Know Your Agent infrastructure, and cryptographically signed audit trails to prove when an agent acted within its mandate.

The Bottom Line

Agentic payments redefine who can participate in commerce.

The winners will be the financial institutions, merchants, and tech platforms that can verify AI agents, make their data machine-readable, and safely shift from manual workflows to autonomous execution.

Once software can act, it will need a wallet. And once it has a wallet, commerce changes.

If you’re building at the intersection of AI, payments, commerce infrastructure, or financial automation, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

This week on Eximius Echo, we’re looking at one of the most important Consumer AI opportunities in India: Health & Fitness.

The First Problem Is Getting Care

The Size of the Market

In a country of 1.47 billion people, healthcare is one of the most fundamental needs to address. AIIMS Delhi alone had 4.8 million OPD patients across 2024-25, pointing to the scale of demand in India’s soon-to-be $320 billion healthcare ecosystem. The core challenge remains access.

The Access Issue

India has only 1.3-1.5 hospital beds per 1,000 people, against a global median of 3. Rural users often travel hundreds of kilometres for basic care. At the same time, India has a growing chronic disease burden, with around 90 million adults living with diabetes and 220 million with hypertension. The problem is both access and long-term wellness.

Tech Penetration in India

Alongside infrastructure investment, technology can help bridge this gap. Healthcare in India has been slow to adopt technology at scale, but recent nationwide initiatives are changing the base layer:

These are important building blocks, but individual user adoption remains limited. Healthcare in India still feels reactive, short-term, and generalised. The next leap will come from intuitive, personalised, and reliable solutions that help Indians access care and manage their health better. This is where AI enters the conversation.

AI is Hastening Innovation

Globally, AI is changing how people interact with health, moving users from one-off symptom checks to deeper guidance on fitness, recovery, sleep, nutrition, and long-term wellbeing. Around 40M users approach ChatGPT daily for health-related questions, including broader queries around wellness and longevity.

While still very nascent in its adoption and evolution, AI is already engendering considerable progress across 3 key domains:

1) Personalised Coaching

AI’s biggest utility lies in its ability to retain long-standing contexts and provide precise answers catered to individual users. Levering this functionality, companies are now creating AI-powered personal healthcare coaches that analyse a user’s entire health history and lifestyle to provide personalised recommendations on improving their overall health. While this can be utilised for short-term ailments, the real application is managing long-standing complications.

2) Enhanced Imaging Diagnostics

Medical imaging is a highly precise process that entails a lot of moving parts. In addition to being expensive, the process frequently causes discomfort and anxiety amongst patients. AI can hasten scans by 75% while producing higher-quality results. Further, due to its ability to account for anomalies that might be imperceptible to the naked eye, AI is also very useful in analysing the results and sharing an initial diagnosis of all the aberrations.

3) Real-time Monitoring

While wearables like wrist-bands have boomed in popularity over the last decade, most of these solutions operate in siloes. AI can not only engender a cohesive ecosystem, tying all the different streams of data together, but it also allows for real-time interface with physicians. This can prospectively also lead to specific solutions for remote monitoring, particularly for elderly patients, to track their vitals and flag any alarms as proactively as possible.

Notably, all of this will accelerate with greater implementation of AI across other facets. While hovering around 15% globally, AI’s addressable market share in healthcare is expected to grow to 30% by 2030.

This shift has also produced companies like Thrive Health, Nori, and Function Health, which are building hyper-personalised AI health concierges around individual health data, guidance, and longevity.

The Bigger Opportunity Is Daily Health Guidance

In India, the AI x Health opportunity is unlikely to be limited to premium wearables, longevity subscriptions, or quantified-self tools built for affluent users. The larger opportunity lies in making health guidance more accessible, contextual, and continuous.

Today, the average healthcare journey is still fragmented. A user visits a doctor, gets a prescription, runs tests, receives lifestyle advice, and is then expected to manage the rest independently. This gap is especially visible across chronic conditions, fitness, nutrition, preventive health, and recovery.

AI can become the layer that connects health data, doctor advice, lifestyle habits, and everyday decisions. The strongest wedges are likely to emerge across three areas:

1) Chronic Disease Companions

AI-led programs can help users manage diabetes, hypertension, PCOS, thyroid, cardiovascular disease, asthma, COPD, and chronic pain through reminders, risk flags, report interpretation, adherence nudges, and doctor escalation.

2) Medical Assistants

Voice-first, multilingual AI assistants can help users in Tier 2+ India navigate symptoms, prescriptions, lab reports, medication, and next steps across low-friction channels like WhatsApp, calls, and lightweight apps.

3) Personalised Health Coaches

India needs online-first health guidance that understands regional diets, languages, family structures, affordability, and local health behaviours.

The real opportunity is to turn fragmented information into simple, reliable, everyday guidance that users can access repeatedly, not just when they are already unwell.

The Real Moat Is Trust

For AI to meaningfully change health in India, access alone will not be enough. Healthcare is deeply personal, high-stakes, and dependent on reliability. Users will adopt AI health tools when they feel accurate, safe, contextual, and easy to act on.

This makes trust the real moat.

The strongest AI x Health companies in India will need large amounts of local health data to train well across symptoms, prescriptions, diagnostics, chronic conditions, regional diets, languages, and patient behaviour. This data depth can help them move from generic advice to guidance that works for Indian users.

Usage can then become a compounding advantage. As users return across symptoms, reports, medication, chronic care, nutrition, fitness, and recovery, the product can retain more context, personalise better, and improve recommendations over time.

This also helps fix a broken LTV/CAC equation in consumer health. Historically, users engage episodically, and platforms pay repeatedly to reacquire them. AI can make health engagement more continuous, increasing retention while reducing servicing cost over time.

The UI/UX also needs to improve. Today, users are stuck between expensive doctor access and unreliable Google searches. AI can create a better front door: conversational, vernacular, personalised, clinically informed, and able to escalate to doctors when needed.

If you’re building at the intersection of AI, healthcare, fitness, chronic care, or personalised health guidance, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

This week on Eximius Echo, we’re looking at one of the most important Consumer AI opportunities in India: Education & Upskilling.

For the last decade, Indian EdTech was largely built around access: recorded classes, test prep apps, cohort-based learning, tutor marketplaces, hybrid coaching centres, and certification platforms.

These models expanded reach, but the core experience often remained similar for every learner: same lesson, same pace, same test, and same dashboard.

AI changes that.

It allows learning to become adaptive, conversational, multilingual, and deeply personalised. More importantly, it allows the system around learning to become smarter: lesson planning, question generation, assessments, student records, doubt-solving, feedback, and content management can all move from manual workflows to intelligent infrastructure.

The question now is simple: ‘Can AI transform Indian EdTech from a content-delivery market into a personalised-learning market?’

THE SCALE IS ALREADY HERE

India’s EdTech market reached $2.8Bn in 2024 and is projected to reach $33.2Bn by 2033, growing at a 28.7% CAGR from 2025 to 2033.

The underlying education system is even larger.

India has 1.5Mn schools, 8.5Mn+ primary and secondary teachers, 260Mn+ school enrolments every year, and 40Mn+ higher education students across 1,000+ universities and 42,000 colleges.

This makes India one of the largest learning ecosystems in the world.

The opportunity is also fundamentally different from global markets. In the US or Europe, AI education may optimise for premium tutors, advanced assessments, and institution-led deployments. In India, the larger wedge is access-led: affordable mobile-first learning, regional-language content, voice interfaces, AI tutors, automated assessments, and job-linked upskilling at scale.

WHAT THE FIRST EDTECH WAVE BUILT

India already has strong education companies across multiple layers:

• PhysicsWallah built one of India’s largest affordable test prep and hybrid coaching platforms.

• Classplus helped tutors and coaching institutes digitise their own content, payments, and student management.

• UpGrad built for higher education and professional upskilling.

• SpeakX is using conversational AI to help users improve spoken English.

That is where the second wave begins: not at the level of more content, but at the level of interface, behaviour, and daily learning workflows.

At Eximius, we are also seeing this shift through our own portfolio. CuePilot is building a voice-first operating system for preschools, using voice as the default interface for attendance, observations, parent communication, report cards, and administrative workflows. Another Eximius portfolio company, currently in stealth, is building an AI-driven consumer learning product for Gen Alpha, a cohort already tuned to audio, video, and dynamic content.

WHERE THE GAP STILL SITS

The biggest gap in Indian education is no longer only access to content. It is access to good guidance.

Most students still need help understanding what to study, how to study, when to revise, where they are weak, what questions to practise, and how to stay consistent.

Most teachers and institutions still need help creating lessons, organising question banks, tracking student progress, managing assessments, and personalising attention at scale.

AI can bridge this gap across three layers:

1. Lesson creation

AI can help teachers and institutions generate lessons, quizzes, assignments, assessments, and supplementary reading material on the fly. This is especially useful for India’s coaching centres and schools, where teachers often operate with limited time and large classrooms.

2. Personalised learning journeys

AI can move learning from fixed curriculum delivery to adaptive paths. It can understand a student’s pace, strengths, weaknesses, language comfort, and exam goals, then act like a personal tutor that creates a plan, guides practice, and changes direction as the student progresses.

3. Content and data management

Coaching centres and institutions sit on massive repositories of question banks, recorded lectures, test papers, assignments, and notes. Much of this remains poorly stored or under-utilised. AI can organise, tag, retrieve, and repurpose this material into revision modules, mock tests, doubt-solving flows, and personalised practice sets.

WHAT COMES NEXT

We see five AI-led shifts shaping Education & Upskilling in India, with affordability becoming the unlock. A private tutor can cost ~₹1,000 per hour. An AI tutor can potentially deliver always-on support at ~₹1,000 per month. For Tier 2 and Tier 3 India, that cost shift can meaningfully expand access to personalised learning.

1. AI-led coaching

AI tutors can become a 24/7 learning layer for doubt-solving, homework help, study planning, and revision. As models improve on reasoning, accuracy, and empathy, AI-led coaching can become a primary support system for students. This is also where our stealth Gen Alpha learning bet fits in, with AI shaping a more dynamic, audio-video native learning experience.

2. Voice-first and multilingual learning

India’s next learning interface may be spoken, not typed. Regional-language voice interfaces can help learners across India 2 and India 3 access tutoring, skill training, and exam support in the language they are most comfortable with.

3. Adaptive assessments

Instead of static tests, AI can generate assessments that adjust in real time. Difficulty levels, question types, feedback, and remedial modules can change based on the learner’s performance.

4. AI microlearning for upskilling

Professionals increasingly need short, practical, personalised learning modules. AI can create micro-courses for interview prep, communication, sales, coding, leadership, domain knowledge, and workplace readiness.

5. AI as institutional infrastructure

The most interesting opportunity may sit beneath the student app. AI can become the operating layer for schools, colleges, coaching centres, and tutors across grading, attendance, lesson planning, student records, assessments, performance tracking, and engagement. CuePilot reflects this shift at the preschool layer, where voice can make institutional workflows easier to adopt and manage.

THE BIGGER CONSUMER AI LENS

Education is one part of a larger shift we are tracking at Eximius.

Consumer AI is forcing every large consumer category to be re-examined from first principles: how we learn, shop, manage health, discover experiences, plan our lives, build relationships, and make decisions.

In education, the direction is already visible.

The next breakout company may not look like a traditional EdTech app. It may look like an AI tutor, a teacher copilot, a coaching-centre OS, a voice-first learning companion, or a full-stack AI academic institution for specific skills.

The strongest founders in this space will understand both sides of the problem: India’s learning gaps and India’s learning behaviours.

If you’re building in AI-led education, upskilling, learning infrastructure, or personalised coaching, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in ConsumerTech, FinTech, and Enterprise AI.

Let’s dive in.

Financial services has always been a hard category for new technology to enter. The systems are older, the data is sensitive, and the cost of failure is brutal.

Yet today, banks, NBFCs, fintechs, and payment companies are paying close attention to AI. The use cases are no longer restricted to chatbots or surface-level automation. AI is entering customer support, fraud detection, credit scoring, KYC workflows, compliance checks, document processing, employee productivity, and portfolio intelligence.

The reason is simple. Financial services runs on decisions. AI can make those decisions faster, cheaper, and more contextual.

But a deeper question emerges: “Can the current infrastructure stack actually support production-grade financial AI?”

THE CLOUD-FIRST DECADE SOLVED THE FIRST PROBLEM

For the last decade, hyperscalers such as AWS, Microsoft Azure, and Google Cloud have become the default infrastructure layer.

They made compute elastic, shifted infrastructure from capex to opex, and gave companies access to storage, security, APIs, identity, analytics, and deployment tools without forcing them to build data centres.

For fintechs, cloud allowed teams to scale during transaction spikes, launch products faster, and focus engineering bandwidth on customer experience instead of hardware maintenance.

But financial AI is starting to stretch the limits of generic cloud infrastructure.

WHY FINANCIAL AI NEEDS A DIFFERENT INFRASTRUCTURE CONVERSATION

AI workloads in financial services are different from traditional SaaS workloads. A fraud system cannot afford latency spikes. A KYC workflow cannot expose sensitive customer data. A payments assistant cannot hallucinate through a transaction dispute.

The infra layer now has to deliver four things together:

This is where hyperscalers start to feel less natural for financial AI.

Shared cloud environments can make performance less predictable. Dollar-linked billing can make costs harder to forecast. And while storing data in India solves one part of the problem, banks still need control over access, governance, auditability, and model deployment. This gap is creating a new category of AI infrastructure.

THE RISE OF SPECIALISED GPU DEPLOYMENT

India-first GPU cloud providers, sovereign cloud providers, and specialised AI infra companies are building environments designed for training, fine-tuning, inference, and regulated deployment.

The market already has early leaders. Neysa is building a sovereign AI execution layer for training, fine-tuning, and production deployment. E2E Networks is making INR-denominated GPU cloud more accessible for Indian AI teams. Yotta’s Shakti Cloud is targeting institutional-scale AI workloads. The whitespace, therefore, is not in basic GPU access alone. It sits in bundled, BFSI-ready infrastructure that combines compute, compliance, security, orchestration, and managed services.

Bare-metal GPU access reduces abstraction. NVLink, InfiniBand, and direct-attached storage improve performance. Rupee-denominated billing and local hosting make cost and compliance easier to manage.

The shift is from generic elasticity to workload-specific performance. That matters because AI in finance will not be one uniform workload.

DIFFERENT FINANCIAL INSTITUTIONS WILL ADOPT DIFFERENTLY

Large banks are likely to follow a hybrid, control-first path. Sensitive workloads like ledgers, credit scoring, fraud detection, and PII-heavy systems will stay on-premise or private, while front-facing channels, chatbots, and marketing analytics can continue on cloud.

Mid-tier banks and NBFCs may become the most aggressive adopters of specialised AI infrastructure. For them, GPU-as-a-service, AI cloud layers, vector databases, RAG systems, and document intelligence tools can unlock faster KYC, loan origination, and customer service workflows.

Small finance banks and cooperative banks may take a more SaaS-led or community-cloud route. The Indian Financial Services Cloud and sovereign AI providers can help them access secure, local AI infrastructure without carrying the full cost of building it internally.

This adoption curve will be segmented. Each customer type will need a different product, price point, and deployment model.

INDIA ALREADY HAS A PROOF POINT

NPCI’s FiMi shows where this market can go.

FiMi is a payments-native AI model built for Indian financial workflows. It supports transaction disputes, mandate lifecycle management, and multilingual interfaces across English, Hindi, Hinglish, Telugu, and Bengali. It powers the UPI Help Assistant and shows that financial AI in India needs domain-specific data, local context, high-volume reliability, and infrastructure that can operate inside regulated payment systems.

This is the larger signal: India’s financial AI stack will need models that understand Indian money movement, Indian languages, Indian compliance, and Indian scale.

THE ECONOMICS WILL DECIDE THE BUSINESS MODEL

Cloud and APIs work well when AI usage is still experimental. But once workloads become predictable and always-on, the cost equation changes.

The lesson is simple: when utilisation is high, infrastructure ownership or specialised deployment can become a major cost advantage.

This matters for financial AI because fraud detection, risk monitoring, KYC processing, customer support, and compliance automation can run continuously. These are the workloads where infrastructure design becomes a margin lever.

WHAT MAKES THIS VC-FUNDABLE

Pure GPU resale is a thin business. The opportunity becomes interesting when a company builds a bundled software and services layer around the hardware.

That could mean GPU orchestration, deployment tooling, monitoring, compliance workflows, inference optimisation, model governance, managed services, or vertical AI stacks for specific BFSI customers.

The strongest companies will show:

In this category, the underwriting question is simple: Can the company turn a capex-heavy infrastructure layer into a high-margin, sticky, recurring platform?

THE NEXT PHASE OF FINANCIAL INTELLIGENCE

The last phase of fintech was application-led. Dashboards, apps, chatbots, digital onboarding, and analytics defined the surface area. The next phase will be infrastructure-led.

As AI becomes embedded inside decisions, financial institutions will need to decide where intelligence lives, how it is governed, and what architecture allows it to scale safely. The winners in this market will combine performance, compliance, and capital efficiency.

In Indian BFSI, AI infrastructure will become a strategic layer. Compute, data, models, and governance will sit at the centre of how financial institutions build trust, reduce cost, and compete in real time.

If you’re building at the intersection of AI, infrastructure, financial services, or regulated workflows, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in ConsumerTech, FinTech, and Enterprise AI.

Let’s dive in.

The last year sharpened the market in a few important ways.

Capital became more selective, not absent. In India, startup funding eased 8% year on year to $11Bn in 2025, even as the ecosystem continued to consolidate around higher-conviction outcomes. Globally, AI moved well beyond fascination and into the centre of venture allocation, accounting for 48% of all venture funding in 2025, the highest share on record. Geopolitics kept redrawing supply chains, incentives, and strategic priorities. In that environment, the companies that kept standing out were the ones solving hard problems with stronger products, clearer use cases, and deeper operating discipline.

That broader shift shaped us, too.

At Eximius, this was the year Fund II took real shape. It was also a year that sharpened our pattern recognition. Fund I outcomes gave us stronger clarity on what early quality actually looks like. Our sector work deepened across ConsumerTech, FinTech, and Enterprise AI. Our community, research, and sourcing engine widened our aperture across the pre-seed market. We now see roughly 800 to 1000 deals a month, which gives us a broad view of where conviction is forming early and where noise still outweighs substance. One thing kept standing out: the earliest stage still rewards clarity before consensus.

Three themes kept coming back to us across the year.

Those themes did not stay theoretical. They shaped where we invested.

9 NEW BETS WE MADE THIS YEAR

1. MyNaksh

An AI-led astrology product that feels less like a marketplace and more like India’s first astrology copilot. What stood out to us was the daily retention loop: a category with deep cultural pull, strong ritual behaviour, and real monetisation potential in a market where even skeptical Gen Z users still end up relying on guidance, pattern recognition, and personalised insight.

2. DevAssure

A horizontal QA platform solving hard testing problems, especially in edge cases and regulated environments like fintech and healthcare. We liked the combination of technical depth, product-native testing, and an India-to-the-world wedge in a category with broad software relevance and growing global demand.

3. Triple Tap Games

A gaming studio building hybrid-casual visual puzzle titles with stronger production value, deeper narrative, and real IP potential. This is a timely bet on a format that AI is making far more viable: visual puzzle games that used to be expensive and difficult to build, now opening up for focused teams with strong taste and disciplined execution.

4. Amicco

A spare-parts marketplace with a moat rooted in operations, not just software. Inventory orchestration, warehouse depth, logistics discipline, delivery within 45 minutes, and 20% margins in a category where the norm is often single digits all matter here. This is what a structurally better marketplace can look like.

5. DecoverAI

A vertical legal AI platform built for lawyers and legal teams, with a strong wedge in commercial disputes. What we liked was the move beyond search toward better insight: more context-aware discovery, stronger legal intelligence, and a product that helps teams work through large, complex case material with more precision.

6. CuePilot

A voice-first operating system for preschools, designed so users do not need to learn a traditional system at all. No dashboards to navigate. No heavy clicking. Just voice as the default interface for attendance, observations, parent communication, report cards, and admin. That product insight felt unusually sharp to us.

7. EarthSync

A techno-commercial intelligence layer for solar, wind, and battery assets. As renewable markets become more complex, spreadsheet-led planning and consultant-heavy decision-making start to break. EarthSync replaces that with a software layer built for forecasting, procurement, and infrastructure precision in markets now shaped by sub-hourly settlement and more dynamic operating logic.

8. Stealth

An AI-native operating system for agribusiness execution, built for the coordination layer that still lives in WhatsApp, calls, photos, and spreadsheets. We were drawn to software that does not just report on activity, but actually helps run the field, reduce human middleware, and bring structure to messy real-world execution.

9. Stealth

A consumer learning product being built for Gen Alpha, a cohort already tuned to audio, video, and more dynamic forms of information consumption than traditional coursework allows. What stood out to us was the behavioural shift itself, along with the strength of the direct-to-consumer subscription opportunity emerging around it.

WHAT THE YEAR TAUGHT US

A few patterns kept repeating across our work this year.

Some of the most compelling early-stage companies are being built in categories where the economic value is already large, but the software is still poor.

The strongest AI companies are not adding intelligence to existing tools. They are turning labour-heavy workflows into software and fragmented processes into systems.

Founder-market fit matters even more when companies are still pre-scale, and the market around them is still taking shape.

And in India, many of the most exciting opportunities still sit in sectors that look messy, operationally hard, or easy to overlook from the outside.

That is often where category-defining businesses begin.

WHAT WE’RE EXCITED ABOUT NEXT

Looking ahead, we’re most excited by businesses where the market has already done the hard part, and the product can now do the differentiating.

Over the last year, we’ve made 9 new bets. Over the next year, we plan to make 10 more. If you’re building for the future from India, for India and the world, across ConsumerTech, FinTech, or Enterprise AI, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. Through this newsletter, we explore the structural shifts shaping tomorrow’s markets.

Let’s dive in.

India’s startup ecosystem has entered a new phase. The conversation is no longer only about how much capital is available. It is about what that capital is actually helping founders build.

As some of you may know, we recently launched the Pre-seed Investment Playbook 2026 in collaboration with 1Lattice. The report was built to understand a market we believe is becoming one of the most important layers in Indian venture capital today: pre-seed, not as a loose category of early cheques, but as the stage where company quality, execution discipline, and long-term graduation potential begin to take shape.

Download the full Playbook here: DOWNLOAD

Here are 5 things you should absolutely know about the pre-seed funding and investing landscape in India.

1. PRE-SEED IS NO LONGER A SIDE LANE. IT IS BECOMING THE STRUCTURAL ENTRY POINT.

One of the clearest signals from the report is this: pre-seed is the only major funding stage that grew year-on-year in 2024-25. India’s pre-seed ecosystem has also expanded roughly 3X since 2020. In a market where later-stage capital has remained cyclical and more selective, this is a meaningful shift.

This tells us something important. Investors are moving earlier, but with sharper intent. Founders are also approaching this stage with greater seriousness. Pre-seed is increasingly where conviction gets underwritten, not only on the idea, but on the founder, the speed of learning, and the path to building something enduring.

2. INDIA’S CHALLENGE IS NO LONGER CAPITAL ACCESS. IT IS GRADUATION.

The headline number from the report is hard to ignore: fewer than 20% of Indian startups reach Series A by Year 4. In the U.S., that number is closer to 50%.

That gap captures the real problem. India has expanded its early capital base through angels, syndicates, family offices, micro-VCs, and accelerators. What has not scaled at the same pace is structured support around product-market fit, go-to-market design, early hiring, governance discipline, and downstream fundraising readiness.

This is exactly why pre-seed matters more today. The first cheque is no longer just catalytic capital. It is a filtering-and-enabling layer. It gives founders time, structure, early feedback loops, and the operating support needed to improve their odds of graduating.

3. THE FOUNDER PROFILE IS GETTING STRONGER, AND THE MARKET IS REWARDING IT.

Nearly 45% of seed-funded founders in India are now second-time entrepreneurs. The report also shows that companies with one or more repeat founders raise materially larger seed rounds than the market average.

That shift is changing the texture of the market. Repeat founders bring sharper judgment, stronger networks, faster validation cycles, and often a more disciplined approach to capital. Investors are responding accordingly.

There is also a broader mindset shift underway. The era of “burn first, figure it out later” has clearly faded. Even at the pre-seed stage, founders are being evaluated on clarity of wedge, speed of execution, monetisation potential, and operational readiness. That raises the bar, but it also makes the ecosystem healthier.

4. INDIA’S NEXT STARTUP WAVE IS BROADENING GEOGRAPHICALLY AND DEEPENING INSTITUTIONALLY.

Nearly 50% of DPIIT-recognised startups now originate outside Tier I hubs. That is a major structural shift. Entrepreneurial ambition is spreading beyond Bengaluru, Delhi, and Mumbai, and the next wave of companies is likely to emerge from a much wider set of cities and founder backgrounds.

At the same time, the investor base is evolving quickly. Micro-VC participation has grown nearly 4X since 2021. India now also has 300+ family offices managing approximately US$30B in AUM, with increasing exposure to early-stage opportunities. Between 2020 and 2025, angel investors deployed roughly US$1.7B across more than 3,000 deals.

Domestic capital is steadily becoming the stabiliser of the ecosystem. That matters. It reduces dependence on global liquidity cycles and gives India’s earliest-stage market a more resilient base.

5. THE NEXT WINNERS WILL NOT BE BUILT ON OPTIMISM ALONE. THEY WILL BE BUILT ON EXECUTION DEPTH.

The report points toward a future shaped by applied AI infrastructure, embedded finance, climate and mobility, space and defence tech, healthtech rails, and commerce and logistics infrastructure. These are large, structurally important categories. They are also categories where execution quality matters early.

That is why “investable” at pre-seed is being redefined. Clean cap tables, ESOP planning, compliance hygiene, MVP or pilot validation, analytics infrastructure, and basic data-room readiness are no longer nice-to-haves. They are signals of institutional readiness.

For founders, this means choosing pre-seed partners who can do more than wire money. For investors, it means moving from intuition-led early bets to more structured conviction. For the ecosystem, it means recognising that the hardest work in venture often begins before traction becomes visible.

India has already built scale. The next chapter will be shaped by how well it builds from zero to one.

Over the last 1 year, we’ve invested in 10 companies. Over the next 1 year, we plan to invest in 10 more.

If you’re building for the future from India, for India and the world, across ConsumerTech, FinTech, or Enterprise AI, write to us at pitches@eximiusvc.com. We’d love to hear what you’re building.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. Through this newsletter, we explore the structural shifts shaping tomorrow’s markets.

Let’s dive in.

For the past few years, most conversations around AI have been shaped by consumer and enterprise use cases: chatbots, coding copilots, productivity tools, image generation, and application development. While that framing still matters, it no longer tells the whole story.

AI is now becoming a core layer of modern warfare.

What began as software built to write code, interpret text, analyse images, and automate workflows is now being deployed in intelligence systems, cyber operations, autonomous drones, battlefield decision-making, and information warfare. And mind you, the shift isn’t theoretical; it is already visible across active conflict zones, defence budgets, and military procurement priorities.

The U.S. Department of Defence (DoD) now manages 685+ active AI projects, and its FY2026 AI request stands at $13.4 billion. Ukraine produced 2 million drones in 2024 and is targeting 5 million in 2025, many of them increasingly shaped by AI-enabled targeting, autonomy, and battlefield coordination. India, too, is moving quickly, with 75+ AI products reportedly operationalised across border intrusion detection, maritime awareness, and battlefield deployments.

This matters because defence is no longer only about superior hardware. It is increasingly about superior software, cleaner data, faster coordination, and better machine-led decision loops. The side that can process information faster, adapt faster, and act faster starts to gain the edge.

That shift is visible across the defence value chain.

AI Across the Defence Value Chain

At one end, AI is transforming ISR, or intelligence, surveillance, and reconnaissance. Systems like Palantir’s Maven Smart System and TITAN compress targeting cycles from hours to minutes by fusing drone, satellite, and battlefield data into a unified picture. Israel’s Gospel reportedly generates 100+ targeting recommendations a day. Ukraine’s Griselda uses NLP to analyse intercepted Russian communications in near real time.

At another end, AI is becoming central to cyber warfare. Offensive systems can scan networks at machine speed, generate malware, and exploit vulnerabilities at a scale no human team can match. Defensive AI is now the only credible response. The U.S. DoD’s FY26 budget includes $14.3 billion for cyberspace activities and $1.6 billion for cryptographic modernisation, underscoring just how central this domain has become.

Then there is the informational battlefield, perhaps the most publicly visible and least understood front in this new AI-led era of warfare.

Deepfakes, synthetic media, and AI-generated propaganda are no longer fringe internet curiosities; they’re strategic tools. Over the past year, we’ve seen manipulated narratives travel at extraordinary speed, especially during moments of geopolitical tension. During the India-Pakistan conflict, deepfake and synthetic war content became a vehicle for projection, fear, confusion, and narrative capture. A deepfake involving journalist Palki Sharma also reflected how trusted public voices can be repurposed to seed misinformation and shape perceptions at scale.

This is what makes AI in defence more expansive than drones and weapons. It is shaping not only what happens on the battlefield, but what civilians, media, governments, and adversaries believe is happening. In that sense, AI is influencing both kinetic warfare and cognitive warfare. That same logic is now reshaping battlefield operations.

AI in Hardware and Battlefield Systems

It is no longer just about unmanned systems. It is about autonomous coordination between systems. Ukraine’s Swarmer has deployed battlefield swarm software in active combat 100+ times. Other companies are building modules that can be attached to ordinary FPV drones, enabling autonomous last-mile targeting even in jammed environments. The next leap is drone swarming, where machines do not simply wait for central instructions but coordinate with one another in real time during deployment. One may defend while another attacks. One may reroute based on threat signals from the other. That turns a fleet into a distributed intelligence system, not just a collection of machines.

And none of this works without the stack beneath it.

The opportunity does not sit only at the application layer. It runs far deeper. Infrastructure like edge AI chips, secure cloud, and communications backbones matter. Middleware that allows fragmented military systems to talk to each other matters. Applications built on top of that stack matter too, whether in intelligence fusion, cyber defence, simulation, or autonomous operations. While your point of entry may be the application layer, the long-term strategic value often compounds lower in the stack, and that’s precisely why this theme matters beyond defence-tech specialists.

It is also showing up in who is winning. For decades, military procurement was dominated by legacy primes built around hardware-heavy programmes and long development cycles. That structure is starting to shift. The Anduril $20B Army contract and Palantir’s $10B Army software contract mark a structural turning point: AI-native companies are now winning at the same scale as Lockheed Martin or Raytheon. That matters beyond defence because it signals where strategic value is beginning to accrue. In sectors shaped by software, data, and machine-speed coordination, the winners may increasingly be the ones building the orchestration layer, not just the asset itself.

Where India Can Build

India, too, has a credible opening here. The opportunity set is already visible across a few distinct fronts:

Policy support is becoming more tangible as well. 430 iDEX contracts have already been signed, 619 Indian startups are engaged, and grants of up to ₹25 crore are available for startups building across AI, autonomy, satellite communications, and cyber tech. Defence exports reached ₹23,622 crore in FY2024-25, with a ₹50,000 crore target by 2029.

Why This Matters Now

We are not a defence fund, and we are not underwriting the next generation of missiles or armoured systems. But modern defence capability is no longer being shaped by hardware alone. It is being shaped by intelligence, coordination, compute, and control. For countries looking to build real strategic depth, AI capability must be built the same way: from applications, through middleware, and down to infrastructure and sovereign systems.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. Through this newsletter, we explore the structural shifts shaping tomorrow’s markets.

Let’s dive in.

Our Annual Day Summit left us with one clean takeaway: the market is rewarding clarity.

Clarity in distribution choices. Clarity in retention loops. Clarity in operating cadence.

Across 2 days, 150 curated attendees, and 10 sessions, we heard the same patterns repeat across capital, public and private markets, and company-building:

1. Founders get rewarded for building with conviction and precision.

2. Investors get rewarded for underwriting what compounds.

That shift shaped our first-ever Annual Day, a closed-door setup intended to pressure-test what matters now, and what the next decade of Indian outcomes will be built upon.

Across multiple sessions: from public markets to private markets, from FinTech to AI workflows, the same idea kept resurfacing - durable companies get built through repeatability. Repeatability shows up in how you acquire, how you retain, how you allocate capital, and how you run the business week after week.

Here are seven signals we’re taking forward, and how founders and investors can use them.

7 signals from the summit

No. 1: Pre-seed has scaled. Graduation stays hard.

Pre-seed in India has grown ~3X since 2020 and stayed resilient through 2024-25. The tougher truth sits one step ahead: fewer than 1 in 5 pre-seed-funded startups reach Series A within four years. In the U.S., this figure is over 50%.

If you’re a founder, treat “graduation” like a product problem. PMF sits at the centre of it. Build your weekly cadence around two questions:

1. “What is the one metric that predicts retention?”

2. “What is the one channel that predicts repeatable distribution?”

If you’re an investor, diligence keeps moving closer to: “how will this company compound?” While market size matters, look for execution systems and a clear path to PMF, as these decide outcomes.

No. 2: Public markets have become a mirror for private companies.

The public markets lens was simple and practical: quality gets measured through predictability. Clean unit economics, reporting hygiene, and steady execution broaden financing options.

Founder takeaway: practice IPO discipline early - monthly closes, clean cohort reporting, tight definitions of retention and contribution margin, and a clear picture of where profitability emerges.

Across public and private markets, returns are generated when one can stay invested.

No. 3: Distribution remains brutal. Precision wins.

This came up in multiple ways - category creation in India-first markets, consumer health scaling, and portfolio journeys. Founders who win treat distribution like a craft: channels, hooks, pricing, and repeat purchase cycles get designed deliberately.

Founder takeaway: pick one distribution wedge and go deep. Every “extra channel” has an opportunity cost. A simple operating rule helps: one wedge, one core promise, one retention loop.

No. 4: Payments infrastructure is mature. The next wins come from what you build on it.

The FinTech panel framed the next decade as “infrastructure to IPO”. UPI created the rails, embedded layers are getting built on top of that, and ultimately, sustainable unit economics decide who reaches scale.

In parallel, AI is transforming all forms of finance - infrastructure, services, and solutions.

Founder takeaway: build with a clear model logic. You can ship fast and still design for sustainability from day one: pricing, risk, and unit economics as first-class decisions.

No. 5: Private markets are segmenting by intent.

The LP-GP conversation made one thing clear: different pools of capital want different outcomes. CVCs, family offices, institutions, and growth investors show up with distinct time horizons and different definitions of value creation.

Founder takeaway: raise from investors whose intent matches your company’s operating reality. Alignment reduces friction and gives you time to build.

No. 6: AI is getting workflow-native.

The AI roadshow kept the focus where it belongs: adoption. AI wins when it saves real time, creates control, and delivers repeatable ROI inside a workflow. Technology has no moat, so the work is adoption and the systems that sustain it.

Founder takeaway: anchor your AI story to one workflow, one user, and one measurable outcome. Trust and governance compound adoption.

No. 7: Long-term capital wants long-term builders.

Across sessions, “long-term” showed up as a discipline: governance, talent density, thoughtful capital allocation, and a focus on compounding.

Founder takeaway: build for durability even while moving fast. The market rewards speed with stability.

A few moments brought these signals to life…

Portfolio in the Spotlight

We also spotlighted companies we’re proud to back: STAN, Cuepilot AI, MyNaksh, EarthSync, OYELA, Triple Tap Games, and DevAssure.

One portfolio moment that landed strongly was STAN’s journey from category creation to category leadership, scaling to $40M ARR in under two years, while executing on a global vision.

The other was the AI Roadshow: three founders building in very different arenas, speaking the same language of adoption. Ankur Agarwal (Cuepilot AI), Devaang Agrawalla (MyNaksh), and Rajat Singh (EarthSync) shared what it takes to make AI stick inside real organisations - clear entry workflows, measurable ROI, and product decisions designed around trust, adoption, and measurable outcomes.

Pre-Seed Investment Playbook (with 1Lattice)

Annual Day also marked the launch of our Pre-Seed Investment Playbook, developed in partnership with 1Lattice, a deep dive into what’s changing in India’s earliest stage.

Download the full Playbook here: DOWNLOAD

To every LP, founder, speaker, and partner who showed up - thank you. The conversations were candid, thoughtful, and deeply constructive. We’re excited to keep sharing what we’re learning through Eximius Echo, and to keep convening exchanges that sharpen conviction, speed up learning, and accelerate compounding outcomes.

Gatherings like this are part of how we’re building Eximius into a long-term institution, compounding trust, learning, and outcomes year after year.

P.S. If you want highlights from the two days, here’s the recap: VIEW

Onward ✨

For years, climate software focused on measurement - counting tonnes, building dashboards, exporting reports. That layer mattered. But as grids tighten, renewables hybridise with storage, and policy becomes more granular, the real bottleneck is no longer reporting. It is decision-making under complexity.

AI is beginning to change that foundation - not by generating more insights, but by turning physics, policy, and capital constraints into optimised actions. The value is shifting from dashboards to decision infrastructure.

If you’re new here, Eximius is a pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. Through this newsletter, we explore the structural shifts shaping tomorrow’s markets.

Let’s dive in.

For most of the last decade, climate software looked like reporting. Count the tonnes. Build a dashboard. Export a PDF. Route it to a sustainability team that sits somewhere between brand and compliance.

That wave mattered, and it built real businesses. But it also trained the market to believe that “climate tech” is mostly measurement plus storytelling.

What we are seeing now feels like infrastructure. The transition is getting physical, regulated, and operational. When grids move to tighter settlement windows, when renewables become hybrids with storage, and when policy varies not just by country but by state, the bottleneck stops being intent and becomes decision-making under complexity.

This is where AI becomes useful, not as a climate chatbot, but as a layer that makes messy workflows legible. The most interesting Climate x AI companies are building decision engines for high-stakes moments: how a project is sized, how a PPA is structured, how an asset is dispatched, how penalties are avoided, and how value is stacked across multiple revenue streams.

The opportunity here is not one product. It is a shift in what the climate stack even means.

The Excel Cliff

Most climate infrastructure decisions still start in spreadsheets. Excel is the default because it is flexible and trusted.

But we are hitting an Excel cliff.

Move from annual averages to 15-minute intervals, and a single year becomes ~35,040 time blocks. A 20-year asset life becomes ~700,800 blocks. Now add multiple assets (solar, wind, batteries), degradation curves, curtailment assumptions, grid constraints, and state-specific charges. Add financing (cashflow, DSCR), contractual clauses (guarantees, penalties), and scenarios (variants to shortlist, then finalise terms). At that point, you do not have a “model”. You have a brittle software product living inside a file, held together by human discipline.

The real problem is not compute. It is governance. When a model becomes too complex to audit, teams simplify assumptions until the model fits the tool. Value leaks quietly. Or decisions start depending on who built the spreadsheet, not what the world actually looks like.

A large share of climate value creation over the next decade will come from preventing that leak.

From Carbon Math to Asset Math

Climate software is moving from accounting to operations: from reporting what happened to deciding what to do next, and what it will do to EBITDA, risk, and compliance.

Renewables are a clean example. Early solar was relatively forgiving. Planning was simpler, grid rules were looser, and contracts tolerated smoothing. The new era is less forgiving. Hybrids and storage introduce control variables. Penalty regimes tighten. Merchant markets and day-ahead participation introduce price signals. Suddenly, the difference between “good” and “great” is not branding. It is math and execution.

In many markets, regulators are compressing the feedback loop. Shorter settlement windows mean errors surface quickly and get priced quickly. Forecasting quality becomes a financial variable, and deviations become a line item.

That is why “asset math” is becoming core. You need software that understands the physics of the asset, the economics of the contract, and the rules of the grid, all at once.

The New Stack: Physics, Policy & Optimisation

Climate is one of the few verticals where physics and policy shape the product at the same time.

Physics is non-negotiable. Solar yields, wind intermittency, battery round-trip efficiency, cycle life, C-rate constraints, and degradation are real. Policy is equally non-negotiable. Grid charges, banking rules, wheeling, net metering, open access, forecasting compliance, deviation penalties, time-of-day tariffs. The same project in two jurisdictions can behave like two different businesses.

The strongest platforms stitch three layers together:

Planning is the Wedge, Operations is the Compounding Loop

In infrastructure, planning is often the easiest wedge because the pain is immediate and the ROI is legible. It is where projects are designed, bids are won, and contracts are shaped.

Small improvements here can be surprisingly valuable. Developers routinely add buffers to be safe, especially in hybrid and storage-heavy systems. If software helps reduce that buffer even slightly, the capex impact is meaningful. Saving 1 MW of unnecessary capacity on a 30 MW project is not a feature. It is money.

But planning alone is rarely the endgame.

Operations is where stickiness compounds. Once an asset is live, the problems become continuous:

The most compelling long-term arc is “an energy desk in software”. A system that understands the asset, the grid, and the contract, and can recommend (and over time automate) actions that protect and improve EBITDA.

Batteries make this especially interesting because they convert uncertainty into control. If you are long, charge. If you are short, discharge. If prices are low, charge from the grid (where allowed). If penalties tighten, optimise for compliance. Value stacking stops being a slide. It becomes a dispatch policy.

Procurement will look like FinTech, Not Consulting

Another theme that keeps resurfacing is procurement, especially for corporates sourcing renewable energy.

A similar dynamic is emerging in Scope 3 programs. Large companies want to help suppliers decarbonise, but last-mile work (tariffs, load profiles, payback math) still lives in spreadsheets and consultants. Standardising that decision surface can beat another emissions report.

Software will not remove relationships, but it can standardise the decision surface.

A good platform does not just list developers. It translates the buyer’s load curve into requirements, normalises submissions into comparable scenarios, and helps the buyer choose based on landed cost, reliability, and risk. Commercially, this often leads to blended models: low-cost planning for the buyer, and a success fee from the winning counterparty aligned with deal size. You see commissions that resemble the offline market (often ~0.75% to 1.5%), but with better transparency and repeatability.

The nuance is that procurement itself is a commodity. Developers win on cost of capital, execution ability, and risk appetite. So defensibility comes from trust and underwriting: delivery history, performance variability, counterparty strength, and sometimes payment default risk. Climate procurement marketplaces will behave more like fintech than e-commerce. The scoring and risk layer will matter as much as the listing layer.

Where Moats Actually Form

“Data moat” is an overused phrase in climate. The durable moats we see tend to be more specific:

And there is a distribution moat: selling through capital. If you can land inside platforms that sit above many assets (portfolio owners, lenders, insurers), you can scale faster than selling project-by-project.

What this means for Founders & Investors

This category is entering a phase where planning and operations start blending. The same engine that sizes an asset can later run it. The same policy layer that informs capex decisions can later drive compliance and dispatch. The same optimisation framework that helps win a tender can later protect margins in the field.

Sequencing is the hard part. Planning is a great wedge, but it can trap you in project services if you do not convert to recurring workflows. Operations is sticky, but integrations can slow early GTM. Procurement is large, but commoditises if you do not own the risk layer.

As early-stage investors, we have found it useful to stay concrete. We care less about “AI for climate” and more about whether the product provably reduces time-to-decision, whether it is credible to both engineers and finance teams, whether policy is treated as product (not a feature), and whether there is a path from insights to actions.

Closing Thought

The climate transition is often described as an energy transformation. True, but incomplete.

It is also a decision transformation. The world is not short of capex. It is short of trustworthy, repeatable decision-making under constraints.

As complexity rises, the winners will not be the companies that shout “AI” the loudest. They will be the companies that turn messy physics and shifting policy into decisions that feel obvious in hindsight, and then turn those decisions into actions that move real electrons, real money, and real outcomes.

If you’re a founder building in this space, we’d love to chat. Reach out to us at pitches@eximiusvc.com

Voice AI has moved from an experimental interface to a foundational execution layer. As 2026 unfolds, Voice AI has become one of the clearest paths for AI systems to move from assistance to action.

Unlike text-based AI, voice operates under real-time, emotional, and trust-sensitive constraints. Conversations are the work. And recent breakthroughs in latency, streaming inference, speech realism, and agent orchestration have crossed a critical threshold: AI systems can now independently handle millions of phone-based interactions that were previously infeasible to automate.

As a result, Voice AI agents are no longer pilots. They are embedded directly into revenue, compliance, and care-critical workflows across enterprises globally, and especially in phone-first markets like India.

But… What is Voice AI?

Voice AI refers to end-to-end systems that can listen, reason, decide, and respond in real time through spoken language, while maintaining conversational context and executing tasks. These systems operate across phone calls, VoIP (Voice over Internet Protocol), and embedded voice channels, and are designed to replicate the dynamics of human conversation, not just respond to commands.

Crucially, Voice AI is not an interface layered on top of an LLM. It is a tightly coupled system where perception, reasoning, and action must occur within strict latency and reliability bounds.

A production-grade Voice AI system includes:

And… Why is Voice AI Its Own Category

• Latency is a hard constraint:

Human conversation breaks beyond a few hundred milliseconds. Voice systems must deliver sub-800ms end-to-end responses and handle interruptions within ~200ms. This requirement reshapes system architecture in ways text AI never encounters.

• Voice carries emotional bandwidth:

Tone, urgency, hesitation, and frustration are embedded in speech. Modern Voice AI adapts in real time, making it uniquely powerful in high-trust domains like healthcare, banking, and collections.

• Conversations are messy:

People interrupt, backtrack, and change intent mid-sentence. Managing overlapping speech and dynamic replanning creates a deep technical moat.

• Unit economics have finally flipped:

Voice interactions are expensive - telephony, real-time inference, streaming TTS. Falling inference costs and better orchestration have crossed the viability threshold, enabling large-scale deployment.

• Risk is higher:

Hallucinations and errors are harder to detect in voice and more damaging in regulated settings. Governance, compliance, and observability are core product requirements, not add-ons.

The Market in 2026

Voice AI Has Moved Into Production, and adoption is strongest where calls are execution, not communication:

• Customer support, cutting Tier-1 volume by 50%+

• Sales and onboarding with higher conversion consistency

• Debt collections and payment reminders with improved recovery

• Healthcare scheduling, intake, and chronic care engagement

• Banking and fintech workflows (KYC, balances, fraud checks)

• Logistics coordination and field operations

• Education, HR screening, and interviews

• Government helplines and citizen services, especially in India

• Consumer companionship and coaching

Funding and Momentum:

Investor conviction accelerated through 2025 into early 2026. Globally, AI investment crossed $1.8B in January 2026, with voice highlighted by a16z as a leading “AI employee” wedge.

High-profile bets reinforced this thesis. Sequoia-backed Sesame raised $250M to pursue expressive, real-time conversational agents, signalling voice as a long-term platform, not a feature.

India saw strong early-stage momentum. In January 2026 alone, Indian Voice AI startups raised $15M+, despite broader funding selectivity. Capital is concentrating around startups delivering 60-70% cost savings versus human Tier-1 support, while meeting regulatory and vernacular requirements.

Notable rounds included Bolna ($6.3M seed, General Catalyst, YC, Blume), building self-serve, multilingual voice agents optimised for Indian telephony; Ringg AI ($5.5M Series A, Arkam Ventures), focused on automating SMB and enterprise voice workflows; and ArrowHead (~$3M), part of a growing cohort building India-first voice infrastructure.

Whitespaces: Where Voice AI Still Breaks

Whitespaces emerge where Voice AI is technically possible but operationally difficult, regulated, or underserved.

Vertical whitespaces -

• BFSI & collections: compliance-native, vernacular agents for high-volume financial workflows. First to pilot, but adoption is still early.

• Healthcare: supervised agents handling non-clinical tasks and follow-ups.

• Logistics & field ops: voice systems resilient to noise and poor connectivity.

• Government services (India): auditable, sovereign Voice AI for massive citizen-facing demand.

• Skilled AI Companions: agents designed to perform specialised, high-trust roles, such as astrologers, fashion stylists, and executive assistants.

User segment whitespaces -

• SMBs priced out of enterprise deployments

• Rural and semi-literate populations where voice is the primary digital interface

Product & infrastructure whitespaces -

• Voice-specific evaluation, QA, and safety tooling

• Voice-first workflow design (not chat retrofits)

• Emotion-aware interaction design

• Compliance, auditability, and governance layers

What It Takes to Win in Voice AI

Winning in Voice AI requires more than models:

• Technical foundations must prioritise low latency, resilient orchestration, and graceful degradation.

• Data moats come from domain-specific, annotated voice data, especially vernacular and code-mixed datasets in India.

• Trust and regulation must be foundational, with audit logs, consent management, and explainable decision paths.

• GTM execution starts narrowly: one high-frequency call type, then expands.

• Ecosystem leverage - telcos, banks, hospitals, BPOs - unlocks distribution and data.

• Teams must combine speech engineering, LLM expertise, telephony, and regulatory fluency.

The Future Is Voice-First Execution

By 2030, Voice AI agents will handle a significant share of routine phone-based work globally. Hybrid human-AI models will dominate sensitive domains, blending efficiency with trust.

Voice AI is no longer an interface. It is becoming infrastructure - a durable execution layer for AI, especially powerful in markets where text-first systems never worked.

The long-term winners will treat voice not as an add-on, but as a first-class system built for human perception, regulation, and real-world execution.

If you’re a founder building in this space, we’d love to chat. Reach out to us at pitches@eximiusvc.com

For decades, commerce scaled by expanding catalogs, optimizing search, and pushing traffic through increasingly crowded funnels. Mobile improved convenience, but discovery, trust, and decision-making remained fragmented. Today, AI is beginning to change that foundation - not by showing consumers more products, but by understanding intent, collapsing discovery time, and actively guiding decisions.

This shift is redefining how demand is created, how supply is surfaced, and where value accrues. The winners won’t be those with the largest inventory, but those that own decision-making at the moment of intent.

If you’re new here, Eximius is a Pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about. Let’s dive in.

Marketplaces have always evolved in tandem with technology to reduce friction in the discovery and purchase of goods. Currency replaced barter, department stores introduced standardised pricing and trust, global supply chains unlocked scale, and the internet removed geographic constraints. Mobile commerce then reset expectations around speed, convenience, and infinite choice.

Across each wave, marketplaces consistently optimised four core levers: discovery, pricing, inventory, and fulfilment.

Today, the next inflection point is AI. Commerce is beginning to shift from search-led, manual browsing toward intelligent, agent-driven shopping.

The Shape of India’s Commerce Opportunity

India is one of the world’s largest consumption markets. Online retail stands at roughly $70B (~10% of total retail) and is growing at ~20% YoY, driven by three structural tailwinds:

• a 300M+ online shopper base,

• deeper Tier-2+ penetration (already ~60% of incremental demand), and

• rising purchase frequency as grocery, fashion, and beauty move online.

Fashion (~30%) is among the fastest-growing large categories, fueled by trend-led demand and rising online comfort.

Within online fashion, GMV and volumes are sharply skewed by price tier. Mass fashion drives ~50% of GMV but ~75% of volumes (low AOV, high commoditization). Aspirational fashion contributes ~35% of GMV with ~20% of volumes, while premium accounts for ~15% of GMV on just ~5% of volumes. This skew highlights a critical insight: future GMV and margin expansion will be driven by curation, trust, and fit, not traffic scale alone.

From Browsing to Buying: The Rise of Intent-Led Consumer Behaviour

Consumer behaviour in commerce is shifting from exploration-heavy, search-led journeys to faster, influence-driven, and increasingly assisted decision-making.

• Discovery time is collapsing as intent per session rises: Average fashion e-commerce journeys have compressed from ~20-30 minutes per purchase to ~8-12 minutes, while recommendation-led sessions convert 1.5-2.0× better than search-only flows.

• Social and video are replacing search as the primary discovery layer: Around 30-40% of Gen Z and millennial shoppers now discover products via short-form video (Instagram Reels, YouTube Shorts) before ever searching. Social-influenced purchases account for ~15-20% of online GMV.

• Assisted shopping is scaling faster than expected: Roughly 25-35% of online shoppers have used some form of AI assistance (chat, recommendations, size guidance). Conversational interfaces deliver 10-20% higher conversion rates and 5-10% lower returns compared to traditional flows.

• Convenience and price sensitivity now coexist: Consumers are willing to pay a 5-10% premium for faster delivery or higher confidence (fit, authenticity), even as price comparison shifts earlier in the journey, compressing brand differentiation at checkout.

Where Today’s Marketplaces Break Down

Fashion discovery isn’t broken because of lack of choice, it’s broken because of lack of confidence, context, and clarity. Discovery is fragmented across Instagram, influencers, and offline cues, while checkout, trust, and fulfillment remain anchored to legacy marketplaces.

At scale, this fragmentation compounds:

• Discovery fails to convert intent: Marketplaces host millions of SKUs, yet ~5% of inventory drives ~80% of GMV. Mass users face price clutter, aspirational shoppers struggle to find differentiated brands, and premium users can’t easily discover niche labels - leading to weak intent capture and inefficient CAC.

• Fit, returns, and trust erode unit economics: Fashion return rates of ~25-40%, driven by poor fit and expectation mismatch, inflate logistics costs and compress margins. Premium buyers also demand authenticity, sizing confidence, and service quality that horizontal marketplaces struggle to deliver consistently.

• Personalization remains shallow: Despite vast data pools, recommendations are largely rules-based and non-contextual. Assortment skews toward commoditized SKUs, while long-tail niches remain underserved - lowering conversion and weakening repeat behaviour.

Demand Is Fragmenting, Not Scaling Uniformly

Fashion spend is shifting toward micro-segments rather than broad categories. Athleisure and casualwear are growing at ~18-20% CAGR, value sneakers show 2× higher online penetration than apparel, and D2C premium fashion is compounding at 25%+, driven by design, drops, and brand affinity rather than discounts.

At the same time, SKU scale on marketplaces has far outpaced usable depth. Leading marketplaces list 1-5M+ fashion SKUs, yet fewer than 10-15% of brands capture meaningful demand, signalling chronic oversupply in commoditized price bands and persistent undersupply in curated niches.

How Marketplaces Will Evolve

Incumbent platforms like Amazon, Flipkart, and Myntra will evolve into AI-enabled marketplaces, not AI-native ones. AI will be layered aggressively to optimize existing systems rather than re-architect how commerce fundamentally works.

What changes:

• AI materially improves search relevance, recommendations, pricing, inventory turns, return prediction, and logistics efficiency.

• Friction across the funnel drops, lowering CAC and lifting conversion.

• Unit economics expand through better demand forecasting and faster inventory cycles.

What doesn’t change:

• Discovery remains SKU-led and reactive to intent, anchored in search, filters, and comparison.

• Supply stays catalog-first and scale-biased, optimized for GMV, margins, and sponsored placement.

• Long-tail and emerging brands continue to face cold-start problems, limited visibility, and pay-to-play discovery.

AI-Native Marketplaces Create New Demand

AI-native marketplaces are built around intelligence, not inventory. The AI layer is not an enhancement to the product-it is the product. Remove the AI, and the marketplace stops functioning.

What fundamentally changes:

• Discovery shifts from browsing to intent expression. Users communicate mood, occasion, context, or taste, while AI interprets soft signals such as lifestyle, aesthetics, and social behavior.

• Catalogs become dynamic. Static brand listings give way to algorithmically assembled assortments - looks, edits, capsules, or personalized drops generated in real time.

• Discovery becomes proactive. Instead of reacting to search queries, AI surfaces “this fits you now,” moving from brand-led merchandising to user-led taste graphs.

How AI-Driven Shopping Will Look

AI is reshaping commerce end-to-end, from discovery to creation to checkout. The core shift is from browsing products to delegating decisions to intelligent systems.

• Fit and confidence move pre-purchase. Visual try-ons, AR, and AI avatars answer “will this suit me?” before checkout. As 3D body and fabric modeling improve, uncertainty drops and return rates fall.

• Outcomes replace products. AI stylists and digital closets recommend complete looks, not single items, using wardrobe data, context, and lifestyle signals. Commerce shifts from selling SKUs to solving occasions.

• Discovery becomes conversational and visual. Search moves from keywords to intent. Users describe, show, or reference what they want, and AI curates options across brands, budgets, and constraints - compressing decision time.

• Supply, pricing, and execution become agentic. Generative AI enables on-demand design and rapid prototyping, while autonomous agents predict demand, optimize inventory, surface deals, and complete checkout across platforms.

Whitespaces to Build In

India’s AI commerce stack is taking shape rapidly - from backend optimization to shopper-facing experiences - while incumbents embed AI across search, payments, and checkout. The next wave will be less about incremental features and more about who owns the agentic layer.

1. Conversational commerce: Over 70% of India’s online shoppers come from non-metro markets, where search and filters add friction. Voice- and chat-led shopping in regional languages offers a more intuitive path to discovery-led purchases.

2. Circular fashion and C2C resale enablement: Resale is gaining momentum as value-conscious consumers, high wardrobe churn, and social acceptance of thrift converge. AI can formalize this behavior through accurate item valuation, quality grading, and demand matching.

3. Unified commerce data: Even at scale, product data remains fragmented across sellers and channels. As AI increasingly guides purchasing decisions, standardized, real-time access to pricing, inventory, and delivery data becomes critical infrastructure.

4. Omnichannel intelligence: Physical touchpoints still influence a large share of fashion purchases. AI can bridge online preference data with store-level inventory via virtual try-ons, in-store kiosks, and scan-to-buy experiences.

From Inventory to Intent

The next phase of commerce will not be won by platforms that show more products, but by those that own decision-making. As AI shifts shopping from browsing to delegation, value moves upstream, from inventory aggregation to intent interpretation.

AI-enabled incumbents will optimize catalogs and logistics. AI-native marketplaces will do something more fundamental: reduce cognitive load, increase confidence, and deliver outcomes instead of options.

In a mobile-first, influence-led market like India, this transition will happen faster than expected. The real opportunity is not building another marketplace, but building the intelligence layer commerce runs on.

If you’re a founder building in this space, we’d love to chat. Reach out to us at pitches@eximiusvc.com

For decades, accounting scaled linearly with people and hours. Software helped, but the core economics remained unchanged. Today, AI is beginning to alter that equation, not by replacing accountants, but by collapsing time inside workflows, making compliance a baseline, and enabling firms to scale capacity and consistency in new ways.

This shift is reshaping how firms buy software, how startups go to market, and where durable value is being created, across vertical AI wedges in B2B and tax-led entry points in B2C. The winners won’t be defined by model sophistication alone, but by how quickly they earn trust and embed into real workflows.

If you’re new here, Eximius is a Pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about. Let’s dive in.

Accounting is a deceptively large services category because it underpins nearly every business outcome that matters: regulatory compliance, cash visibility, and decision-making.

The US represents a mature, highly formalised market, defined by a long tail of solo CPAs and small-to-mid-sized firms. Tooling adoption is relatively advanced, and workflows are already platform-heavy, even if fragmented across multiple systems.

India, by contrast, is far more fragmented, shaped by a massive SME/MSME base where trust, compliance, and relationship-led delivery still drive buying decisions. At the same time, the scale of demand is unmistakable: the Big Four alone reported ~$4.7Bn in India revenue in FY24, with projections exceeding $5.4Bn+ in FY25 - highlighting both market size and structural complexity.

What tools are typically used?

Accounting stacks are rarely a single system. Instead, they are stitched together across a few core layers. At the base sits the system of record: in the US, platforms like QuickBooks (SMB), NetSuite (mid-market), and enterprise ERPs dominate; in India, Tally remains the default, alongside a growing cloud ecosystem led by Zoho and SMB-focused ERPs.

On top of this, firms are increasingly automating repeatable backbone work - AP/AR (Accounts Payable/Accounts Receivable), month-end close, expense processing, procurement, supplier management, and reporting - because time saved here compounds every month and quarter.

Most AI adoption enters through narrow, high-frequency pain points, including:

• Invoice processing and AP

• AR follow-ups and collections

• Expense categorisation

• Month-end close support

• Compliance checks and anomaly detection

India, in particular, is well-positioned for AI-first service delivery in regulated professional services. Traditional accounting scales linearly with billable hours; AI breaks that constraint, enabling structurally different firms rather than incremental efficiency gains.

How AI can change Accounting Services

While software adoption has increased, core workflows and economics remain largely unchanged. AI adoption can materially change how firms across the board function.

Net effect: The old model monetised time. The new model augments workflows. And that shift is what unlocks the next phase of accounting services - deep advisory.

How much have AIxAccounting Companies Raised Globally?

In the US alone, startups building in the AIxAccounting space have raised over $2Bn. Startups are building in all sorts of accounting practices:

Overall, Bookkeeping, Practice Management and Process Automation have raised over 82% of the total funding. This opens up white-spaces for startups looking to build in Audit, Tax Filing, and Specialty Tax.

Emerging Opportunities We Are Excited About

At Eximius, we are actively exploring AI x Accounting opportunities across enterprise and consumer segments, where GTM innovation will define the winners.

1. Vertical Category Specific Wedges

We are particularly excited by founders who avoid serving “accounting” as a broad horizontal and instead start with a single messy category where bookkeeping and compliance are uniquely painful. This can include AI-led accounting platforms purpose-built for industries such as e-commerce, restaurants, healthcare clinics, and logistics.

Vertical focus enables:

• Pre-built workflows and templates

• Deep integrations with industry-specific systems

• Clear ROI narratives tied to that vertical’s pain points

This enables startups to get access fast, given how niche the accounting practices are for the industries. Solo accountants and mid-size firms can leverage this the most to scale non-linearly and compete with larger players. Eventually, they will need to evolve and add horizontal services to be able to capture TAM.

2. Direct-to-Consumer (B2C) Tax and Personal Finance

We are also excited about consumer tax as a high-intent entry point into AI-led financial platforms. Tax filing is time-sensitive, compliance-driven, and anxiety-inducing, creating strong willingness to pay and a natural wedge for AI-assisted workflows.

The primary challenge with this offering is the frequency of use. Tax filing is typically annual, which limits engagement and lifetime value unless expanded into adjacent journeys.

Successful B2C strategies therefore will layer ongoing financial use cases on top of tax:

• Personal finance and budgeting

• Investment and wealth management

• Credit and lending

• Transition from individual to small business

This model has been proven in mature markets. TurboTax offers multiple tiers, including “Expert Assist” and “Expert Full Service”, blending software with human expertise in a “done with you” or “done for you” funnel. In India, ClearTax explicitly markets AI-assisted filing with an in-product assistant guiding users through returns. However, none of these are breakout examples and we’re excited to talk to founders building in this space.

The Next Phase of Accounting Services

What matters to us in evaluating founders building in this space is not only focus, but also speed of trust, hence a category specific + Workflow specific GTM is what we think will win. We also think that products that start with efficiency as a wedge but immediately prove additional revenue expansion opportunities will win trust at scale.

In closed, referral-driven communities like CPA firms, brand matters disproportionately. We believe vertical specificity increases the likelihood of becoming a “household name” within that niche, making subsequent expansion meaningfully easier.

Finally, our conviction is that GTM will matter more than technology over time. As models and infrastructure continue to commoditize, defensibility will increasingly come from distribution, trust, and brand. Accounting is a relationship-driven category with long sales cycles and high switching costs. Founders who are GTM-first, deeply understand how accountants buy, and design their product around real workflow adoption will outperform teams that lead purely with technical sophistication.

If you’re a founder building in the AI and Accounting Services space, we’d love to chat. Reach out to us at pitches@eximiusvc.com.

Welcome to our 24^th and final newsletter of 2025. Throughout the year, we’ve done our best to introduce you to some of the most important trends and pivotal happenings of the year through Eximius Echo. Before we dive into this edition, we wanted to take a moment to thank all of you for reading every single piece and providing critical feedback that helped shape this endeavour. We look forward to hearing more from you across 2026 and beyond!

In our last edition (check it out if you haven’t), we referred to 2025 as a landmark year. For us at Eximius Ventures, it meant turning a new chapter with the launch of our second fund. While it’s hard to believe that 2021 and the launch of Eximus Fund I are so distant, our deep interactions with founders and operators over the last half a decade have propelled us on our path and fueled our ambition to be the first cheque that India can trust.

As we bring the year to a close, we wanted to talk about what made 2025 special for us. From a slew of exciting investments to marquee events and riveting IPs, here’s a quick snapshot of 2025 from us at Eximius Ventures.

If you’re new here, we’re a Pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about.

Let’s dive in.

New Investments

Over the course of the year, we evaluated around 11,000 companies, spoke with nearly 2,000 founders, and backed 8 new companies - 6 you’ll meet below, and 2 currently in stealth (more on those early next year!)

Here’s a quick snapshot of the new teams we welcomed to the Eximius Family this year:

a) Triple Tap Games (ConsumerTech)

A studio building hybrid-casual hidden-object games that stay easy to enter while still rewarding progression and narrative depth. Built by a founder with prior exit experience and deep genre intuition, the team brings repeat execution muscle and a precise read on where hybrid-casual is headed next.

b) Decover AI (Enterprise AI)

A legal-tech search engine for e-discovery, research, and knowledge management, building toward full-stack legal intelligence. Founded by operators who’ve built and scaled enterprise search systems before, with a firsthand understanding of where legal workflows break and how to deliver immediate ROI, in demanding markets like the US.

c) Amicco (ConsumerTech)

A vertical marketplace simplifying spare-parts procurement for independent garages.

Built by a founder who has lived the problem end-to-end, bringing operator empathy and on-ground insight to a daily, high-frequency workflow where reliability and speed directly impact livelihoods.

d) DevAssure (Enterprise AI)

An AI-augmented test creation and maintenance suite that automates QA through intelligent agents. Founded by engineers who’ve shipped at scale and felt the pain of brittle testing firsthand, with a clear focus on embedding directly into developer workflows to unlock faster releases without regressions.

e) MyNaksh (f.k.a. AstroCulture) (ConsumerTech)

An AI-driven astrology platform designed for personalised, ongoing insights that compound into deeper user relationships. Led by a consumer-tech founding team that understands both scale and sensitivity, building trust-first experiences in a category where belief, retention, and product depth matter more than transactions.

f) CuePilot (Enterprise AI)

A voice-first, agentic preschool operating system unifying parent communication, teacher workflows, curriculum delivery, and fee collection. Built by founders who have run preschools themselves and engineered ML systems at scale, combining lived operational pain with deep technical capability and early proof across live, renewing centres.

New investments mark beginnings. Follow-on rounds mark something quieter, but deeper: trust that compounds. As we welcomed new founders into Fund II, we also watched several from our earliest chapters step into their next phase of growth.

From First Cheques to Next Chapters

Also worth celebrating: members of the Fund I family - Skydo, STAN, Handpickd, Finarkein, 30 Sundays (among others; yet to be announced) - took meaningful next steps this year by raising follow-on capital.

• Skydo raised $10M, with new investor Susquehanna Asia Venture Capital joining the cap table,

• STAN raised $10.5M, with new investors Sony Innovation Fund and Hyderabad Angels Fund (HAF VC) joining the cap table,

• Handpickd raised $15M, with new investors Bertelsmann India Investments and Titan Capital Winners Fund joining the cap table,

• Finarkein raised $1.5M, with new investor DSP Group Family Office joining the cap table,

• 30 Sundays raised $770K, with new investor InfoEdge Ventures joining the cap table, etc.

Seeing founders we backed early continue to find belief, momentum, and partners along the way reinforces why we do what we do.

Beyond capital, 2025 was also about creating spaces for founders and operators to learn from one another, share context, and build alongside the ecosystem. That intent shaped how we showed up, across the year.

New Community Initiatives

‘Community’ is a key part of our identity. With an ecosystem as young and diverse as India, there is never a shortage of exceptional people to learn from, both for investors and operators. This ethos is reflected across some of the marquee events we conducted in 2025:

a) Gen Alpha: The New Consumers: To dissect Gen Alpha as consumers, we convened an esteemed group of operators across the spectrum to shed light on them, and what they portend for the future.

b) Operator Events: We brought together some of the most stellar operators across the Delhi-NCR and Bangalore ecosystems to create a dedicated forum to connect and share insights.

b) The Gaming Mixer: Organized in partnership with Gamedev India, AWS, Bagrry’s, and Blitz, ‘The Gaming Mixer’ saw over 50 developers from across the Delhi-NCR come together to bond over games and India’s potential as a gaming hub.

c) Cities Connect: We set out on a multi-city journey in partnership with MIXI Global Investments and WeWork Labs, covering Gurgaon, Pune, and Hyderabad to speak with some of the most eminent local stalwarts to decipher future trends and what India’s building.

d) Pitch to a Partner: To bridge the gap between founders and investors, we launched the ‘Pitch to a Partner’ program, where founders could solicit crisp feedback from Pearl Agarwal, our Founder and Managing Partner.

In addition to the above, we collaborated with Nvidia for their Inception Program, which empowers Indian builders to embrace the transformative power of AI, and Microsoft for their Accelerate Together event to discuss the innovations that are redefining the landscape and how to best support founders in this new paradigm.

Stepping Stones

Through all these initiatives, our persistent objective is to be as close to the nucleus of the founder ecosystem. As a testament to this ambition, we are proud to announce the launch of our new office in Bengaluru.

Importantly, we also welcomed new members this year on our path to becoming India’s premier first cheque:

1) Aryan Sinha: Analyst - ConsumerTech

2) Sidhartha Menon: Sr. Analyst - FinTech

3) Anusha Desai: Investor Relations & Capital Formation

4) Neha Mourya: Investor Relations & Capital Formation

5) Jyothi: Executive Assistant

6) Sumanta Sengupta - Advisor

We’re looking forward to you meeting them!

To end the year and this piece, we want to take a moment to talk about what 2025 meant to each of us, what inspires us for 2026, and share some stellar recommendations to satiate your curiosity this holiday season.

Humans of Eximius

Partners

Pearl

• What content inspired you in 2025?

Atomic Habits by James Clear was a brilliant read that stayed with me.

• What are you looking forward to in 2026?

Building meaningful relationships that go beyond one deal or one transaction.

Preeti

• What content inspired you in 2025?

Horowitz’s core lesson: leaders don’t get the luxury of certainty. You act anyway, and you own the consequences.

• What are you looking forward to in 2026?

In the next year, I’ll try and invest attention where it better compounds outcomes in all aspects - work and otherwise.

Investments Team

Shubham

• What content inspired you in 2025?

3blue1brown and Veritasium kept the curious child scientist in me alive.

• What are you looking forward to in 2026?

Travelling farther, reading wider, and continuing to develop as a person.

Laksh

• What content inspired you in 2025?

The Outsider by Albert Camus was a great read. I was also moved by the poetry of Faiz, Ghalib, Firaq, and Iqbal. The last of these I hope to read more of in the next year, in particular.

• What are you looking forward to in 2026?

Giving back to founders and operators. I want to help them in reaching their goals and getting to the next level. On a personal note, I want to resuscitate my passion for writing and finally finish The Witcher 3.

Keshav

• What content inspired you in 2025?

The Compound Effect and The Prophet shaped much of my thinking this year.

• What are you looking forward to in 2026?

In 2026, I’m looking forward to more travel, stronger outcomes at work, and steady progress toward a healthier, fitter me.

Sidhartha

• What content inspired you in 2025?

Two books that shaped me were “The Almanack of Charlie Munger” for its mental models and “Letters from a Stoic” for its timeless guidance on living well.

• What are you looking forward to in 2026?

I want to strengthen my network while getting fitter and staying consistent with reading. I also want to finally start my vinyl record collection.

Aryan

• What content inspired you in 2025?

As a huge fan of podcasts, I kept returning to ‘Acquired’ and ‘Founders’ to sharpen my understanding of markets and long-term value creation.

• What are you looking forward to in 2026?

In 2026, the focus is to deepen early founder relationships and build stronger conviction around emerging consumer trends.

Platforms Team

Neha

• What are you looking forward to in 2026?

In 2026, I’m aiming for PRs across the board - in the gym and in the boardroom.

Lokesh

• What content inspired you in 2025?

The Knowledge Project podcast shaped much of my thinking this year.

• What are you looking forward to in 2026?

In 2026, I’m excited to take on bigger challenges at work, travel to new places, and play more sports.

Aastha

• What content inspired you in 2025?

I am referring to my life as a piece of content here, tbh :P - My biggest takeaway was realising that action beats waiting every single time: decisive action always supersedes passivity.

• What are you looking forward to in 2026?

In 2026, I want to bring focus back to personal capital: fuelling my body, mind & soul.

Anusha

• What content inspired you in 2025?

“Trust” by Francis Fukuyama consistently informed how I thought about long-term partnerships, credibility, and institutional relationships.

• What are you looking forward to in 2026?

I’m excited to develop stronger proprietary insights, elevate our LP engagement, and play a more proactive role in fundraising strategy and fund storytelling.

Tanishk

• What content inspired you in 2025?

Superman; all the symbolism behind who he is, of course, but more importantly, the fact that James Gunn restarted the DC Universe. A fresh start is always on the cards.

• What are you looking forward to in 2026?

To be more present every day. Time’s passing by too fast, and that’s something that I’ve not been able to synthesise yet.

Before the Next Chapter

As this year draws to a close, we’re reminded that while capital and content shape how we show up, it’s the people and the conversations that give our work meaning. Eximius Echo began as a simple attempt to make sense of an increasingly noisy ecosystem, to slow things down, add context, and create space for honest reflection at the very earliest stages of building.

In 2025, that intent deepened. Through The Backstage Pass and our broader writing, we tried to spotlight founders not just for what they built, but for the questions they asked, the trade-offs they made, and the resilience it took to keep going.

To all our readers, we hope the days ahead bring you rest, warmth, and time with the people who matter most.

Wishing you a Merry Christmas, a joyful New Year, and a peaceful holiday season.

We’ll see you in 2026.

- Team Eximius Ventures

2025 wasn’t a year of flashy exuberance; it was a year of disciplined building, sharper market fit, and sectoral divergence.

FinTech regained conviction, ConsumerTech found new product-market pockets, Enterprise AI accelerated on genuine enterprise demand, and frontier themes finally moved from hype to deployment.

Early-stage investing stayed resilient, even as later-stage capital tightened. Certain companies broke out, others recalibrated, and a few defined entirely new categories.

If you’re new here, Eximius is a Pre-seed VC fund backing bold ideas in FinTech, ConsumerTech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about.

Let’s dive in.

At the onset of 2025, had we been asked to describe the Indian ecosystem, two adjectives would have sprung to mind – ‘nascent, and exciting’. Now, near the end of 2025, we feel compelled to append the list with one more – ‘matured’.

While still brimming with innovative ideas and path-breaking revelations, the onset of a new wave of diligent and experienced founders has added a great deal of stability to the ecosystem. Instead of chasing popular global trends, we see founders now taking great pains to validate their theses before going gung-ho. This discipline has greatly boosted investor confidence as well, resulting in $12.1 billion invested across 2025 so far (a 39% YoY increase). Early-stage remains the most exciting frontier with 61% of the funding.

Alongside new ventures, the incumbent companies have continued to operate efficiently and reach key milestones. 2025 has seen 13 new-age tech companies going public so far, a once-a-generation inflection point. Strong public market interest for companies like Meesho, Groww, Urban Company, and Smartworks stands as a testament to how far we’ve come in a relatively short period.

2025 was also a pivotal year for us at Eximius Ventures. With Finarkein, Stan, and Shuru successfully raising their next rounds, our portfolio made us proud. Additionally, we also invested in 8 exciting new companies who are all set to propel the ecosystem forward.

As the year winds down, we wanted to undertake a sector-wise dissection of how each sector fared, and some of the most exciting trends that saw taking shape.

Where the Action Was: Sector-by-Sector in 2025

Beneath the topline numbers, three sectors defined where founders built most boldly and investors bet most confidently.

FINTECH

FinTech remained the most popular sector with 30 new tech-led bets announced by VCs in 2025. Projected to reach $250 billion in revenue by 2030, the sector is also heavily leveraging AI across key functions to optimise operations. As per Moody’s, 18% of Indian fintech companies are actively leveraging AI, compared to the average of 9% across all surveyed sectors. This penchant for leveraging frontier-tech has led to the Indian ecosystem housing the third-most fintech firms (>14,500) and the fourth-most unicorns (31).

Key Trends:

i) AI x FinOps: In lending, AI is improving due diligence and underwriting for both consumer and MSME segments, while in fraud, it is identifying risks earlier and automating pre- and post-incident reporting. Insurance players are using AI to streamline claims management that once depended on slow, expensive BPO operations. Even compliance is becoming continuous and automated, freeing teams to apply their expertise to higher-value judgment.

ii) Central Bank Digital Currencies: CBDCs represent the next generation of programmable, government-backed digital money that can streamline how value moves across the financial system. Programmability can enable faster, cleaner, and fully auditable transactions.

iii) AI x Personal Finance: AI-powered systems can deliver unbiased, goal-aligned advice based on a person’s real financial condition and needs. This unlocks deeper financial inclusion, better access to credit, and a more financially aware population.

iv) NRI Wealth Management: There is a clear opportunity for a unified, multi-currency, multi-jurisdiction platform that helps NRIs consolidate, diversify, and optimise their portfolios according to their risk profile.

If FinTech sharpened trust and infrastructure, ConsumerTech shaped how India consumes, learns, and spends online.

CONSUMER-TECH

29 ConsumerTech deals were announced in 2025, trailing fintech by just 1. With over 1.2 billion mobile users and over 86% households having internet access, India represents one of the largest consumer markets in the world, which is expected to reach $300Bn by 2027. Notably, India is also set to witness a digital economy revolution, with the digital spend per capita for India slated to reach $2000 annually by 2030.

Key Trends:

i) AI-led Content Creation: Spending over 47 hours per week online, India is one of the biggest content markets in the world. From interacting with their favourite content creators to consuming material from popular OTT platforms, India features a ravenous user base. Therefore, both individual content creators and production houses are aggressively leveraging AI in their workflows to produce multimedia content as quickly as possible. AI tools are being leveraged across the workflow from ideation to editing.

ii) Virtual Social Ecosystems: 43% of Indians in urban regions feel lonely. Therefore, companies are using AI to create not just companions, but entire virtual friend-groups that can engage with people on their favourite subjects and help them feel better. Additionally, these can also be leveraged for special functions like interview prepping, etc., to give them clear and intuitive use-cases.

iii) Verticalized Quick Commerce: Around 60% of buyers in India now expect same-day deliveries, while 30% prefer options under two hours. While we have already seen the onset of horizontal options like Blinkit, Zepto, etc., there is a large market for offering a distinguished selection and quick deliveries across individual markets like baby-care, BPC, healthcare items, etc.

iv) Personalised Learning Modules: India’s online learning market is expected to reach $1.06 trillion by FY28. Using AI, there is considerable scope in not only aiding young students with their curricula, but also helping adults learn new skills, languages, etc., in a more personalised format. All skilled services can be democratised and productised using AI.

If ConsumerTech reflected how India spends and socialises, Enterprise AI reflected how India builds and operates.

ENTERPRISE AI

In India, Enterprise AI is still nascent but scaling fast; the market was about USD 1.1B in 2024, and is forecast to reach nearly USD 9.9B by 2030, implying roughly a 45% annual growth rate. With 18 deals announced in 2025, Enterprise AI is one of the most exciting sectors in the ecosystem. The industry is witnessing a key shift from isolated automation units to connected, AI-native workflows that enhance reliability, speed and decision making. The strongest companies will be the ones that treat robotics, supply chain systems and operational data as one integrated software problem while focusing on repeatable ROI, rapid deployment, and products that improve as they gather more data.

Key Trends:

(i) AI-led Industrial Automation: Robotics is evolving into a software-first ecosystem. AI-powered perception and planning are raising accuracy to near-perfect levels, reducing cycle times and enabling automation in environments that were previously too variable.

(ii) Supply Chain SaaS: Supply chains are shifting from reactive coordination to anticipatory control. Real-time visibility, AI-led planning and intelligent routing are becoming the new baseline while demand engines, automated exception handling and dynamic load balancing are delivering double-digit gains in efficiency.

(iii) Physical AI Driving Operations: Autonomous mobile robots, drones and exoskeletons are already exponentially improving throughput, quality and safety for warehouses, plants and inspection environments.

Taken together, these sectoral shifts tell a clear story: India didn’t just grow in 2025 - it evolved.

Where We Go From Here

2025 was a first-of-its-kind hallmark year for India. With greater maturity and stability across the spectrum, we are witnessing a generational inflection point in real-time, i.e. domestic capital coming in with greater awareness. However, as exciting as this year was, we see several key trends across major sectors take shape that make us more excited for 2026 and beyond.

If you’re a founder looking to build in these exciting times, please reach out to us at pitches@eximiusvc.com.

As industries battle hazardous workplaces, shrinking labor pools, and rising operational complexity, humanoids are stepping into real workloads - from warehouses and factories to elder-care and inspection. What once lived in research labs is now entering early deployment cycles, backed by breakthroughs in embodied AI, low-cost actuation, and coordinated government support.

But beneath the excitement lies a stark truth: India accounts for less than 2% of the global humanoid market. The next decade won’t be about “building cool robots”, it’ll be about engineering reliability, safety, and scale into a category poised to become the next general-purpose automation platform.

If you’re new here, Eximius is a Pre-seed VC fund backing bold ideas in Fintech, Consumertech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about. Let’s dive in.

Humanoids are no longer a sci-fi curiosity, they’re emerging as a serious response to dangerous workplaces, aging populations, and shrinking labor pools. From hazmat sites to warehouses and elder-care facilities, deployment has already begun. Last year alone, the US recorded 5,283 worker deaths from occupational injuries, while China released 35+ humanoid models in 2024, backed by USD 20B in coordinated state funding. The global humanoid market stands at USD 3B in 2025, with projections of USD 38B by 2035. India, however, accounts for under 2% of this market, and risks falling further behind unless it accelerates policy, capital, and ecosystem development now.

“Humanoid robotics will not only build the biggest businesses we’ve ever seen - by many multiples - but also the most important product launch of our lifetime.”

- Brett Adcock, Founder & CEO, Figure AI

Why Humanoids Are Needed

The rise of humanoids isn’t hype - it’s the consequence of five structural pressures converging globally:

1) Hazardous & High-Risk Workplaces

Over 2.3 million people die annually from work-related causes. Chemical plants, nuclear sites, mines, and disaster zones require human-like reach, balance, and tool handling. Humanoids can step directly into environments built for human ergonomics - no costly infrastructure redesign required.

2) Healthcare & Elder-Care Shortages

By 2030, 1M US nurses will retire; India’s elderly population will reach 150M by 2035. Social humanoids like Pepper and Grace already support emotional care, patient engagement, and monitoring, filling gaps traditional robots cannot.

3) Warehousing Complexity & E-commerce Growth

Warehouses face extreme churn and unpredictable demand. Humanoids can lift irregular items, navigate crowded aisles, and perform repetitive or hazardous tasks without changing existing layouts. Early pilots show 25-30% efficiency gains.

4) Manufacturing Labor Gaps

Factories struggle to automate tasks requiring human dexterity. Humanoids can handle machine tending, small-parts assembly, and quality checks by slotting directly into human workflows, unlocking automation without redesigning production lines.

5) Construction & Outdoor Risk Zones

Construction is one of the world’s deadliest sectors. Humanoids can climb, balance, and operate in unstable terrains, making them ideal for inspection, demolition planning, and hazardous material handling - areas where wheeled robots fail.

What Exactly Is a Humanoid?

A humanoid robot mirrors human proportions and movements to operate seamlessly in environments designed for people. Most models:

• Stand 1.3-2.0 meters tall

• Weigh 20-100 kilograms

• Walk bipedally

• Use dexterous multi-fingered hands

• Carry sensor stacks (cameras, LiDAR, depth sensors, tactile feedback)

• Run 4-24 hours on a single charge

What sets humanoids apart is their general-purpose capability - reach, balance, mobility, and manipulation, all in a form factor that fits into existing human environments. This allows them to perform tasks across factories, warehouses, hospitals, and hazardous zones without remodeling the environment.

The Humanoid Market Today

The market today is spread across industrial automation, healthcare, logistics, social robotics, defence and research, with industrial and elder-care use cases driving the earliest commercial traction.

Estimates vary widely depending on definitions, but the most defensible mid-case places the global market at ~$3B in 2025.

“China isn’t just ahead - it has built the entire humanoid ecosystem end-to-end. From high-precision motors and advanced actuation systems to government-backed labs and robotics talent, the country now produces over 35 humanoid models a year and has scaled orders 10× in twelve months.”

- Sathya Narayanan, Co-Founder, Control One

Use Cases Live in the Market Today

The humanoid ecosystem today is led by a concentrated group of deep-tech giants and fast-scaling robotics startups. What separates them is not just funding, but engineering depth - actuators, control systems, dexterity - and the strength of enterprise partnerships that convert prototypes into real deployments.

Tesla, Figure AI, Boston Dynamics, Unitree, 1X Technologies, and Apptronik now anchor the global race. Together, they’re piloting humanoids across factories, logistics hubs, inspection sites, and early home-assist environments, setting the performance and cost benchmarks the rest of the industry must match.

Investor Outlook: What Will Actually Make Humanoids Succeed

Investors are no longer asking if humanoids will work, they’re asking what must be true for them to scale. Capital today views humanoids as the next general-purpose automation platform, but real success depends on an entire stack maturing at once.

What investors are really underwriting now is ecosystem readiness, not just robot capability. Three layers matter most:

1) Hardware + Intelligence Must Cross the Reliability Threshold

Humanoids need industrial-grade actuators, sensors, batteries, and whole-body control systems that can operate safely in unpredictable environments. Embodied AI - perception, planning, and manipulation - must be stable enough to handle real workloads, not just lab demos.

2) Manufacturing & Cost Curves Need to Bend Fast

Scaling requires specialized production lines, robust component supply chains, and modular designs that push per-unit costs closer to industrial automation. Investors are watching which companies can reliably produce hundreds to thousands of units, not tens.

3) Safety, Certification & Operations Must Be Enterprise-Ready

Standardized testing, compliance frameworks, fleet management, remote monitoring, teleoperation, maintenance networks, these are now make-or-break factors. Enterprises will only deploy humanoids when reliability, liability, and uptime are guaranteed.

The investor view is simple: Humanoids will scale only when hardware, software, manufacturing, safety, operations, ROI, and policy maturity converge.

Government Programs Accelerating Humanoids

Around the world, governments are treating humanoids as strategic infrastructure, not experiments. China is scaling fastest with coordinated industrial policy and over USD 20B in funding that enabled 35+ humanoid models in 2024. The US leads deep-tech and defense-driven R&D through DARPA, NASA, and NSF, powering breakthroughs in locomotion and control. Europe is shaping global safety and compliance standards through the AI Act and Horizon programmes, while Japan continues to push the frontier of balance control and human-safe robotics. South Korea brings manufacturing strength, rapidly iterating in actuators and hardware. India has strong intent and early frameworks through initiatives like the National Robotics Strategy, IndiaAI Mission, and PLI incentives, but must accelerate funding, supply-chain depth, and execution to stay competitive in the global humanoid race.

How India Can Catch Up

“India’s robot density is just 7 robots per 10,000 workers - far below the global average of 162 and China’s 470. To catch up, India must focus less on building every hardware component and more on training the best available robots for sector-specific applications. The real value will lie in the deployment and service layer.”

- Sathya Narayanan, Co-Founder, Control One

India sits at a strategic inflection point. With coordinated policy, capital, and talent, the country can build a globally competitive humanoid ecosystem by 2030. The roadmap spans three horizons:

• Immediate (6-12 months): Set up a National Humanoid Robotics Task Force, deploy ~70 humanoids across hospitals and hazardous industries to create real demand signals, and introduce a Robotics Talent Visa to attract global expertise.

• Medium Term (1-2 years): Establish three Centers of Excellence focused on software, actuators, and sensors, and launch a PLI-style incentive to localize critical robotics components.

• Long Term (3-5 years): Create a USD 500M India Humanoid Innovation Fund to back domestic champions and roll out a national safety + certification framework aligned with global standards.

In total, India will require USD 650-1,000M over five years, alongside private capital, to secure a meaningful position in the global humanoid race.

Where the Humanoid Race Goes from Here

Humanoids have moved from sci-fi speculation to strategic infrastructure. With breakthroughs in AI control, sensing, actuation, and coordinated government support, the global race is now underway, and the next five years will determine which countries lead.

Between 2026 and 2030, a few hard indicators will reveal whether humanoids are scaling into a high-volume automation category or remain confined to pilots:

• Commercial shipment volumes: showing whether manufacturing bottlenecks are easing.

• Cost-per-unit decline: convergence toward industrial automation pricing will be the strongest readiness signal.

• Government procurement: early public-sector deployment in healthcare, disaster response, defense and hazardous industries will anchor domestic supply chains.

• Safety & regulatory certifications: opening doors to hospitals, factories and public spaces.

• Real-world deployment density: humanoids actively working in logistics, manufacturing, inspection and elder-care, not staged demos.

Globally, China is scaling fastest, the US leads deep-tech robotics, Europe sets safety standards, Japan contributes decades of research leadership, and South Korea strengthens hardware manufacturing. India has the ambition and early policy architecture, but must accelerate capital, supply-chain depth, and regulatory execution to stay in the race.

The next half-decade will define the global hierarchy of humanoid robotics. Countries that align capital + policy + research + industry + talent will capture the economic and strategic advantage of what may become the most transformative general-purpose machine of the coming decade.

If you’re building in frontier robotics or embodied AI, we’d love to hear from you: pitches@eximiusvc.com

As pet ownership rises and homes humanize their furry companions, India’s pet industry is entering a new phase. What was once an unstructured market of home-cooked meals is now seeing the rise of fresh food subscriptions, protein-first diets, and premium, functional nutrition.

But beneath the boom lies a gap - commercial pet food still accounts for less than 10% of total calories consumed. The next wave won’t just be about feeding pets better - it’ll be about building trust, science, and scalable distribution into every bite.

If you’re new here, Eximius is a Pre-seed VC fund backing bold ideas in Fintech, Consumertech, and Enterprise AI. We use this newsletter to share insights, trends, and ideas from the sectors we’re passionate about. Let’s dive in.

India’s pet industry is expanding rapidly, driven by rising pet ownership, urbanization, and the growing humanization of pets. Dogs dominate the market, while cats are the fastest-growing segment. Within this, pet food is the largest and most dynamic category, capturing the bulk of spending and serving as the entry point for most first-time pet parents. Distribution has moved beyond traditional retail to modern trade, quick commerce, and D2C platforms, supported by higher disposable incomes and greater awareness of pet nutrition.

Yet, despite double-digit growth, India remains significantly underpenetrated compared to global benchmarks. This creates a large whitespace for innovation in premium, functional, and personalized nutrition, from breed- and age-specific formulations to supplements and fresh food formats.

The Pet Economy at a Turning Point

India’s pet population has reached nearly 32 million (around 28 million dogs and 4 million cats) with two-thirds concentrated in urban centers. Growth is driven by millennials and Gen Z, especially single-person households and young professionals with rising disposable incomes.

Attitudes are shifting fast: over 70% of urban pet owners now see pets as family, making nutrition and health central priorities. While home-cooked meals still dominate, awareness around balanced diets, breed-specific nutrition, and premium food formats is accelerating.

Dogs consume roughly 3% and cats about 4% of their body weight daily, underscoring the long-term potential for branded food across dry, wet, and treat categories.

Pet parents today fall broadly into four behavioral archetypes:

1. Value Buyers: Prioritize affordability; rely on home-cooked meals.

2. Brand-Loyal Deal Seekers: Trust established brands but chase discounts.

3. Convenience-First Consumers: Choose diet-specific food, shop online, and value time.

4. Experience-Oriented Shoppers: Seek personalized, wellness-led, and premium offerings.

Sizing Up the Pack

India’s pet industry is valued at $5.58B (2025) and projected to reach $12.8B by 2030, growing at a robust ~18% CAGR. Rising ownership, urban affluence, and the shift toward premium, health-focused products are fueling this growth.

Within the ecosystem:

• Pet food drives ~45% of market value. Based on India’s urban pet base and consumption patterns, the TAM for pet food (dogs + cats) is estimated at ~$2.5B.

• Healthcare makes up ~40%, covering veterinary care, preventive health, and wellness products worth $2.21B.

• The remaining ~15% spans grooming, boarding, training, and accessories, reflecting a more lifestyle-driven pet economy worth ~$850M.

How Much Are Indians Spending on Their Pets?

Urban pet parents in India now spend an estimated ₹50,000 annually on pet care, roughly 5-8% of household income.

Here’s how that breaks down:

• Healthcare (50%): vaccinations, vet consultations, and preventive care.

• Food (15%): despite being the largest category by market size, its share in spending remains modest due to reliance on home-cooked meals.

• Grooming (15%): driven by the growing adoption of professional services.

• Accessories, daycare & walking (30%): reflecting lifestyle-driven consumption.

Spending trends signal a clear premiumization wave, led by urban millennials and Gen Z who view pets as family and prioritize health, convenience, and quality experiences.

Pet Food: A $2.5B Opportunity, Just 20% Tapped

India’s pet stack is broad and food remains the largest subcategory by market value, yet one of the least penetrated. Against a ~$2.5B industry, only about ~$0.5B (or 20%) is actually spent on pet food. The gap stems from deeply ingrained habits of home feeding, limited availability of affordable options, and low awareness about balanced pet nutrition.

Roughly 65% of pet owners still feed a mix of homemade and packaged meals, while only 19% rely on commercial food. Dogs account for nearly 88% of India’s pets, but remain largely home-fed; in contrast, cats, just 8% of the population, are far more dependent on packaged food, making cat nutrition one of the fastest-growing subcategories.

As the market matures, with rising awareness of balanced diets, emergence of curated and functional formats, and convenience-led purchasing behaviors, the pet food category is poised to compound much faster, closing the penetration gap and expanding well beyond its current share.

Who’s Shaping India’s Pet Ecosystem

Mars International India with brands like Pedigree, Royal Canin, Whiskas, Eukanuba, Sheba, and Chappi, along with the Indian Broiler Group, which owns Drools and Purepet, together account for roughly 70% of India’s pet food market.

As demand surges, the aisle is crowding with legacy giants and value brands now competing alongside new D2C players that combine products with subscriptions, fresh formats, and complementary services.

Players in Pet Food:

Segmentation by Format:

The Shift Beyond Kibble

While home-cooked meals remain the dominant feeding practice, awareness around balanced diets, breed-specific nutrition, and premium formats is steadily rising, creating headroom for branded players. Yet, commercial pet food, especially for dogs remains deeply underpenetrated in terms of calorie share, underscoring how early the market still is.

At the same time, consumer behaviour is shifting. Many pet parents are moving away from traditional kibble toward fresh meat, home-cooked meals, and alternative feeding formats, driven by three key factors:

1. Health and Nutrition Awareness: Rising focus on pet wellness is prompting owners to question the nutritional adequacy of mass-produced dry food. Fresh, natural, and minimally processed diets are increasingly perceived as healthier alternatives.

2. Transparency and Ingredient Control: Fresh or home-prepared meals offer visibility into ingredients, aligning with owners’ dietary values and fostering greater trust.

3. Erosion of Trust in Legacy Brands: Quality concerns, ranging from stale or contaminated food to inconsistent packaging, have dented confidence in established mass-market labels.

Where the Next Wave of Growth Lies

These evolving preferences are driving demand for fresh food subscriptions, customized diets, functional treats, and high-protein alternatives, opening new whitespace for innovation:

• Human-Grade Meals:

Urban pet parents increasingly aspire to feed fresh, home-style, human-grade meals but lack the time to prepare them daily. A structured D2C subscription model can bridge this gap, offering convenience without compromising health. Players like Heads up for Tails’ “Sara’s” line have begun testing the space, but the market remains fragmented and underpenetrated.

• Protein-Focused Foods & Supplements:

Adequate protein intake is critical for pet health, yet most Indian brands under-deliver, relying heavily on rice and grain fillers. Dogs require roughly 2.5-3.5g of protein per kg of body weight daily, while cats need 4-5g per kg. Most packaged foods, and vegetarian home diets, fall short, creating an opportunity for functional, protein-first formulations.

• Raw & Minimally Processed Diets:

Globally, the raw and freeze-dried segment has become a multi-billion-dollar niche, led by brands like Stella & Chewy’s, Instinct, and Primal Pet Foods. Driven by the belief that such diets mimic pets’ ancestral eating patterns, this trend is now catching on in India, presenting room for premium entrants offering tested, safe, and convenient raw-food solutions.

Beyond the Bowl: Emerging Whitespace Opportunities

Outside of pet food, several underpenetrated yet fast-emerging categories are taking shape in India’s pet ecosystem, driven by rising health awareness and a growing focus on holistic pet care:

• Vet Telemedicine & Diagnostics

With fewer than 20,000 practicing vets for over 32 million pets, access remains the industry’s biggest bottleneck. A hybrid model combining teleconsults with at-home diagnostic collection can bridge availability and trust gaps. Video consults, doorstep sampling, and unified digital health records can create a continuous care loop, where owning diagnostics and data effectively means owning the pet health stack.

• Pet Insurance

Penetration remains below 1%, even as medical costs climb with rising pet humanization. The opportunity lies in micro-premium, API-first insurance embedded across vet clinics, telehealth

platforms, and breeder networks. Instant settlements and preventive features, like wellness credits and vaccination reminders, can transform insurance from reactive protection to proactive care.

• Pet Health Wearables

Next-gen collars are evolving beyond GPS tracking to monitor vitals like heart rate, temperature, and activity levels, flagging issues before symptoms appear. The real moat isn’t hardware, but the biometric data layer, powering predictive vet care, personalized nutrition, and insurance scoring. A closed-loop ecosystem connecting diagnostics, teleconsults, and wearables could redefine pet care as preventive, data-driven, and continuous.

Winning India’s Pet Market: What Founders Must Solve For

India’s pet care boom is real, but growth alone won’t guarantee success. For new entrants, winning means solving for trust, focus, and operational edge in a market still fragmented and price-sensitive.

1. Brand Trust Is the Real Moat: This category runs on emotion, not impulse. Transparency, nutritional credibility, and vet-backed validation will matter more than influencer buzz. In pet care, trust is the true moat.

2. Mass Is Crowded, Premium Is Open: Mass and mass-premium food segments are already dense. The next opportunity lies in functional, fresh, and high-protein formats, where storytelling and science can command pricing power.

3. Optimize for Shelf Life and Speed: Fresh and functional formats demand cold-chain readiness and optimized shelf life. Balancing long-life SKUs for modern trade with short-cycle products for D2C and Q-commerce will separate serious players from opportunists.

4. Build Through High-Quality Contract Manufacturing: Contract manufacturing with certified partners allows flexibility, speed, and scale, but only if quality control and traceability are built in from day one.

5. Position with Precision: Winning brands will pick a lane - science-first, human-grade, or health-focused - and defend it with clear, consistent positioning that builds habit, not hype.

As India’s pet care economy matures, the next generation of founders won’t win by mimicking global incumbents, they’ll win by localizing trust, premiumizing value, and operationalising empathy.

At Eximius, we’re tracking founders building the future of India’s pet economy, from nutrition to health to tech. If you’re one of them, write to us at pitches@eximiusvc.com