Louis-François here from Towards AI with the biggest thing we’ve shipped since our courses.

Towards AI Mentorship, an engineering support program that puts 15 senior AI engineers behind your production and career decisions.

Getting your work reviewed by senior engineers, building the judgment for production decisions, making sure your resume carries the right details - that’s how people actually succeed in this field, and it takes time with the right people. Companies hire senior teams or bring us in. Individuals hire consultants at $200–500/hour. Most people have access to neither.

We’re giving you ours. Our entire team.

See what’s inside the mentorship →

The core of this membership is simple: you always have a senior engineer behind you. Someone who looks at your system and tells you what holds and what breaks before you ship. Someone who reviews your resume and tells you specifically why you’re getting filtered out. Someone who builds production thinking with you: the tradeoffs, the judgment calls, the engineering decisions that a demo will never surface, and that hiring managers are actually screening for.

Post your question any day. Get a written answer from an engineer who’s deployed that system. A real diagnosis from someone who’s been in production long enough to see what you can’t yet.

Everything else is built around that core:

• Live Q&A twice a week, across two time zones: bring your architecture tradeoff, your stuck deployment, your career question. You’ll always find a session that works for you.

• Every month, you get what we call a production blueprint. Best practices we’ve learned from deployment work, or a full system we’ve already built for clients, handed over with the judgment calls behind it. The first one: our current setup and best practices for working with agents as engineers, the same setup and tips our team uses daily.

• Resume, LinkedIn, and portfolio reviews: written feedback in 48–72 hours from engineers who’ve been on the hiring side.

• Our Python for LLMs and LLM Fundamentals courses, included from day one. Every other course: 25% off. Courses in development: 50-90% off at alpha.

• Monthly workshops with industry leaders from Google, OpenAI, and senior engineers in the field.

The engineers behind it: Omar, production LLM systems, AI Engineer World’s Fair. Fabio, one of the most-read NLP newsletters in the field. Samridhi, AI scribe serving 200+ physicians across 105 facilities. Louie, ex-VP J.P. Morgan, Towards AI co-founder. Plus 11 more. Not marketplace mentors. Our team.

We launched this week, and we’re being upfront about that. There’s no wall of testimonials yet. What there is: a team that’s been answering these questions in Discord, in course channels, and in DMs for years and genuinely enjoyed doing it. We just gave it structure. The first people who join will shape this alongside us, the same way our earliest students shaped the courses. We’d love for you to be one of them.

Join for $99/month →

$899/year saves 24% · 30-day money-back on yearly · cancel monthly anytime

— Louis-François, Co-founder at Towards AI

Louis-François here from Towards AI, and I’m excited to share that I keynoted AI Engineer World’s Fair and presented a system and I have been building for the past 18 months that turns thousands of personal notes into persistent agent memory that actually compounds over time.

AI Engineer World’s Fair is where the biggest names in AI engineering share what they’re building, OpenAI, Anthropic, Google DeepMind, Cursor, Hugging Face, and this year, we were right there with them.

And we’re excited to bring you front-row access to the whole thing.

The recording is premiering right now on the AI Engineer YouTube channel. You can watch the whole thing for free.

Watch “Turn 10,994 Notes Into Memory”

What the talk covers:

If you’ve worked with Claude, Codex, or any coding agent, you know the problem: you have thousands of notes, highlights, repos, and PDFs across tools like Obsidian, Readwise, Notion, and Google Drive, but none of it carries over between sessions. Every conversation starts from zero and loses everything when it ends.

We built a research wiki that agents maintain for you. It grows with every session instead of resetting. No vector database, no knowledge graph, nothing to host, just Markdown, YAML, and folders.

In the talk you’ll see:

— The full evolution from V1 (barely worked) to V3 (actually stuck)

— Four Claude Code skills you can install today

— Two live demos pulling in GitHub repos and URLs in real time

— The full open-source codebase (MIT)

Watch the keynote

We’re also at the World’s Fair in person this Sunday.

Samridhi Vaid, Omar Solano, and I are running an 80-minute hands-on workshop: “Context Engineering in 2026: Compaction, Memory & Cost.”

A different angle from the keynote: how to keep agent context lean and affordable as you scale. How to compact, what to keep, what to drop.

If you’re at the conference or in San Francisco, come say hi, I will be there all day. DM me.

📅 2:20 pm–4:20 pm, June 29 · Moscone West

👉 ai.engineer/worldsfair/schedule

The keynote is live right now! Go watch it and send it to someone who’d get something out of it.

Watch “Turn 10,994 Notes Into Memory”

—

P.S. We also ran a 2-hour workshop at AIE London on building multi-agent systems with MCP servers, recording and code are live too. Watch it here.

Good morning!

This is Louis-François again from Towards AI.

I wanted to share something I’ve been working on for a little while for current and upcoming AI engineers:

My full Start AI Engineering in 2026 roadmap.

You can find it here: https://github.com/louisfb01/start-ai-engineering

This is basically the AI engineering version of my old Start Machine Learning repo dating from 2021, but updated for what the field actually looks like now.

And the field changed a lot from when I first started getting into AI in 2018.

Learning AI engineering in 2026 is not just “learn prompting” or “build a chatbot.”

It’s learning how to make decisions around:

• when to use prompting, RAG, fine-tuning, workflows, or agents

• when not to use an LLM at all

• how to evaluate outputs

• how to debug failures

• how to trace systems

• how to deploy and monitor them

• how to use coding agents without outsourcing your thinking

That last point is important.

Codex, Claude Code, Cursor, Gemini CLI, and all these tools make it easier than ever to build something that looks impressive.

But building faster is not the same as understanding what you built.

And this is the main reason I made the repo.

I wanted a guide for people who want to become real AI engineers, not just people who can ship fragile AI demos with an agent.

The repo includes:

• beginner Python resources

• foundational AI and LLM videos

• books I recommend

• free and paid courses

• RAG, agents, evals, MCP, deployment, safety, and coding-agent resources

• project ideas

• communities

• newsletters

• people to follow

• job-search advice

• difficulty levels from 1️⃣ to 🔟 for each resource

I also added a suggested path so you don’t just stare at 300 links wondering where to start, like most LinkedIn posts and listicles out there.

The rough flow is (optionally):

1. Learn the basics and vocabulary (usually with YouTube videos).

2. Pick one free course and one framework.

3. Pick one or two books for stronger foundations.

4. Optionally take one or two applied courses with real projects.

5. Build small systems that break. (repeat)

6. Add evals, tracing, and deployment before pretending it is production-ready.

OR, build your custom curriculum by giving the repo link to Claude/Chat and telling it your learning preferences. Nobody learns the same way! I added a prompt inside the repo that you can paste into your AI agent to turn the roadmap into a personalized learning plan based on your background, time, budget, and goals.

That’s the real learning.

Not just watching videos. Not just collecting bookmarks. Not just asking an agent to build everything.

Building, breaking, fixing, and understanding why.

So yes, you can use AI to learn AI engineering.

You probably should.

But the goal is still to build your own judgment.

Here’s the repo again:

https://github.com/louisfb01/start-ai-engineering

If you find it useful, please star it and share it with one person who wants to get into AI engineering this year.

That helps a lot and tells me whether I should keep improving it.

Hope you find this useful!

Speak soon (hopefully this Sunday!),

p.s. If you see a great resource missing, feel free to reply to this email with it. I want this to become one of the most useful starting points for AI engineers in 2026.

If you’ve ever used AI to write an email, a blog post, or a project update and felt like you spent more time editing the output than it would have taken to write it yourself, chances are your draft looks something like this:

• It opens with “In today’s rapidly evolving landscape.”

• You keep removing “delve,” “tapestry,” and “it’s worth noting.”

• There are enough em dashes to fill a novel.

• The content is accurate, but it reads like it could have been written by anyone, about anything.

• You publish it anyway because the deadline won’t wait.

We dealt with the exact same thing for over three years at Towards AI, editing, rewriting, and occasionally questioning our life choices. Eventually, we decided to stop fixing drafts one at a time. We made one cheatsheet for the entire team to use every time they generate content, so the slop gets caught in the prompt before anyone has to read it.

Today we’re sharing it with our community for free, partly because if we have to read one more ‘devle’ and see another em dash, someone on the team is going to snap.

The Anti-Slop AI Writing Guide is a prompt template with 50+ banned words, style rules, and structural constraints baked in. You paste it into ChatGPT, Claude, or whatever LLM you use, fill in your topic and audience, and the AI follows your rules instead of making up its own. We’ve used it for emails, blog posts, reports, proposals, scripts, and it holds up across all of them. No technical skills, no setup, just copy, paste, and stop editing the same problems out of every draft.

Get the Anti-Slop Cheatsheet (Free)

The guide teaches you how to:

• Give the AI your outline, your section order, and your paragraph rules so it stops defaulting to listicles and generic five-part blog structure

• Ban specific sentence patterns that are AI fingerprints, not just words like “delve” but structures like “It isn’t just X, it’s Y” and openings like “In today’s fast-paced world.”

• Set accuracy guardrails so the AI doesn’t overstate claims, fabricate certainty, or ignore your source material

• Use a second AI as an editor that audits the draft against your anti-slop rules and flags what to fix, so your own edit is a final pass, not a rewrite

It is designed to move the cleanup process into the prompt itself and provides a two-model AI framework to speed up your editing workflow. Download the guide, fill in your topic, and let the prompt do what you’ve been doing manually.

Download it free here

Louis here from Towards AI, where we have spent years building LLM systems with one goal: sharing what we build and, more importantly, what we learn building it, so you can grow as an AI engineer without hitting every wall we did.

Part of that is done through our courses. But the bigger part is making your actual building process easier, every day.

So we decided to share the markdown files we use internally, the ones you can feed directly into an agent. It builds on the context that usually takes years to develop.

Access everything on our repo here: https://github.com/louisfb01/ai-engineering-cheatsheets

Most of the variance between a good product and a poor one isn’t the model, the architecture, or the prompt. It’s what your agent knows before it starts. Two teams with identical setups will produce completely different results if one has loaded the right decisions up front and the other hasn’t. Any agent’s defaults are generic and broad by design. These files replace those defaults with judgments built on real systems.

What’s Inside Our 3 Markdown Files

Most teams reach for multi-agent systems the moment a task feels complex. But complexity isn’t the signal; unpredictability is. Before you commit to an architecture, ask three questions:

• Can you map the steps in advance? If yes, a workflow will outperform an agent every time; it’s faster, cheaper, and deterministic.

• Does the task require decisions that depend on intermediate results you can’t anticipate? That’s where a single agent earns its place.

• Does it require genuinely parallel workstreams that operate independently across different domains? That’s the only case where multi-agent adds more than it costs.

Most tasks that feel like they need three agents need one. Most tasks that feel like they need one agent need a workflow. Our agentic AI engineering cheatsheet gives you the exact questions to ask, the signals to look for, and the tradeoffs we’ve seen play out on real systems.

The other two markdowns are:

Our AI Engineering Playbook, which includes tables for model selection, prompting strategy, RAG setup, memory patterns, eval methods, production config, and much more.

Our anti-slop AI writing guide with our 7-section prompt template, banned-word list, style rules, and our two-model write-then-review workflow. It allows us to make LLM outputs not sound generated.

These 3 files come directly from the Towards AI Academy courses, the same frameworks we teach in depth, distilled into references you can use today. No course required. No paywall. No email catch or anything.

Just a repo you can clone and use: https://github.com/louisfb01/ai-engineering-cheatsheets

If you want to go deeper, full lessons, code, and hands-on projects, that’s what the Towards AI Academy is for.

Louis here from Towards AI. Here’s one thing I keep seeing everywhere:

2025 gave us agent hype. It didn’t give us a reliable way to build them. Most developers are still guessing: which tools to use, how to wire the system, and how to catch failures with evals and monitoring before users do.

So after nine months of building, breaking, rebuilding, and stress-testing, Agentic AI Engineering is finally live. Our newest course, built together with , is designed to teach you how to design, build, evaluate, and deploy autonomous AI systems.

See what you’ll build (syllabus + projects)

Here’s what early students said after going through the material:

“Excellent in depth handling of tradeoffs in evaluating and deploying agent based solutions. A useful mixture of theory and practice, learnt the hard way by expert practitioners.” — Cathal Curtin

“Every AI Engineer needs course like that.” — Ahmed Medhat

“Industry-focused, emphasizing real-world constraints rather than flashy demos, and highly hands-on.” — Abreham Melese

What You Will Build

In the course, you’ll build two agent systems and learn how to keep them reliable when the environment stops being friendly: when tools fail, inputs get messy, latency matters, and “it worked once” isn’t useful.

You’ll build a Research Agent that runs iterative loops, integrates real tools, produces structured artifacts, and supports human-in-the-loop checkpoints with clear stopping conditions. Then you’ll build a Writing Workflow Agent that turns that research into structured, multi-modal outputs using evaluator–optimizer patterns, orchestration, versioning, and state.

But the core of the course is the reliability layer most agent content skips: you’ll design eval datasets, implement LLM judges and pass/fail checks, add observability with tracing, and set up monitoring so you can debug regressions quickly and improve the system deliberately, rather than guessing.

Check out the full course details →

Who Is This For?

This is engineering-heavy and opinionated, designed for developers who want depth. You’ll feel at home if you’re comfortable with Python + LLM APIs, have basic cloud familiarity, and don’t mind debugging failures that aren’t clean.

We built the course by starting with a system we’d actually use, pushing it until it broke, then turning those failure modes into the curriculum, refined through 180 alpha testers. The goal is to prepare you for what agents are judged on in 2026: operational reliability—measurable quality, inspectable behavior, and controlled autonomy.

If your goal is to build systems that survive production and the AI era, start here.

The early-bird seats sold out in under a week. The next 100 seats are now $499 (the lowest available price after early bird). You get lifetime access, ongoing updates, Discord access, live introductory calls, and a 30-day refund if you go through the early material and realize it’s not what you need.

Get access now →

Louis here from Towards AI. We’ve been working on something I’m genuinely excited to share with you today.

We’re launching a FREE Agentic AI Engineering Guide: 6 Mistakes Developers Make When Building Agents—a 6-day email course created with .

After 3+ years building production agentic systems, we kept seeing the same failure patterns break system after system. This course distills those production failures into the exact patterns separating demos from reliable systems.

Here’s how it works:

Sign up free → Get Lesson #1 immediately → One lesson daily for 6 days → Apply to your systems as you learn

If you’ve experienced any of these:

• Agents that work in demos but drift in production

• Changes feel risky, and you can’t predict what breaks

• Costs spike with no clear explanation

• Infinite loops and random decisions

• Every release needs slow manual QA

This course shows you exactly how to fix them.

Get your first lesson now (free)!

What you’ll learn over 6 days:

Mistake #1: Why treating context windows as unlimited buffers destroys reliability, and how to manage your most scarce resource

Mistake #2: Why complexity keeps you from shipping and the simple-first approach that works

Mistake #3: When agents make systems fragile vs when workflows outperform

Mistake #4: Why regex parsing creates time bombs and how structured outputs create reliability

Mistake #5: What separates real agents from naive tool loops (hint: embedded planning)

Mistake #6: How to build evaluation-first systems that catch regressions before users do

What’s inside every lesson:

Each day, you get a complete breakdown of one critical mistake:

• The failure pattern: See exactly how this breaks production systems (with real examples from our builds)

• Why it happens: Understand the root cause so you can spot it in your own systems

• The proven fix: Get the exact solution we use in production, ready to apply immediately

By Day 6, you’ll transform how you build:

• Reduce costs by 4-15x through strategic context window management

• Ship faster by choosing workflows vs agents vs hybrids based on your actual use case

• Eliminate random behavior with structured outputs instead of fragile text parsing

• Build reliable agent loops with embedded planning that’s goal-directed, not reactive

• Deploy with confidence using evals as tests to catch regressions before users do

• Diagnose failures instantly by knowing exactly which of the 6 mistakes is causing issues

These aren’t theoretical concepts. They’re the exact decisions that separate engineers who ship reliable agentic systems from those stuck debugging random behavior.

Start the free course (first lesson in 2 minutes) →

Louis-François here, from Towards AI, the team that turns enterprise AI pipelines into practical training you can actually use.

Lately, we’ve been watching a lot of smart teams ship a slick agent demo in a day… and then quietly suffer for weeks. Tool chaos. Weird loops. Latency creep. A cloud bill that feels like betrayal. And the worst part is you can’t “prompt” your way out of it, because the problem usually isn’t the model. It’s the architecture decision you made when everything still looked easy.

So we built what we wish existed before anyone ships an agent: the Agent Architecture Cheatsheet + a 1-hour webinar that makes the workflow vs single-agent vs multi-agent decision repeatable, across any product, any industry, any team.

It is officially out today, free, and intentionally short.

Download it for free and unlock the webinar access!

“Agent” has become a default setting, like the correct answer to every problem is “LLM + tools + loop + hope.” It demos beautifully… until production walks in and asks the only questions that matter: what’s the failure mode, where’s validation, what state are you persisting, why is it calling seven tools, and why did cost triple overnight?

Most of the time, the best system is embarrassingly unsexy: a workflow with clear steps, sharp edges, and guardrails that don’t depend on the model having a good day. In production, nobody cares if it’s “agentic.” They care if it’s reliable, fast, and doesn’t set money on fire.

That’s what this cheatsheet does for you. It gives you a fast, repeatable way to choose the minimum viable complexity: Workflow → Single Agent + Tools → Multi-Agent. You run a quick autonomy test, answer 12 high-signal questions, and your architecture choice stops being a debate. You’ll know what to build, what to keep deterministic, when autonomy actually buys capability, and when you’re about to pay multi-agent costs for a workflow problem.

And the questions aren’t fluff, they’re the ones that decide whether your system ships: tool sprawl (and when it starts to degrade performance), validation patterns (hard checks vs judge-based), state + persistence, human-in-the-loop gates, latency and cost budgets, and the minimum eval + tracing you need so you can iterate with signal instead of vibes.

This isn’t theory content. It’s the same framework style we use when designing and deploying systems under real constraints, work associated with teams at Thinkific and Europol, because those environments don’t tolerate “it passed the demo.”

If you’re building an agent right now, do yourself a favor: download it, run the decision framework once on your current idea, and let it tell you what to build next, before your prototype turns into a rewrite.

Grab it here for free!

That’s what our monthly cohort kick-off is designed to solve.

This Sunday, the January cohort kicks off with a live session led by our CEO, Louie Peters. The session walks through how learners actually use Towards AI to move forward: how they choose a path, what to work on first, and how people go from concepts to real projects instead of staying stuck in tutorials.

If you joined recently, this is your on-ramp.
If you’re still exploring and deciding whether to join, you’re welcome to attend and see how the learning paths work before committing.

By the end of January, some learners will be:

• building AI apps in Python instead of just reading about them

• working with prompting, RAG, agents, and evaluation in structured projects

• shipping systems they can explain, demo, and build on

Others will still be bookmarking posts and waiting for the “right time.”

The cohort kick-off happens once a month, and January’s session is this Sunday.

👉 Join the January cohort kick-off here: https://calendly.com/taipartnerships/llm-developer-course-cohort

The session is open to anyone exploring or newly joining Towards AI.

Louis-François here, from Towards AI, the team that turns enterprise AI pipelines into practical training you can actually use.

This holiday season, I want to give you something more useful than inspiration.

Because “I’m going to do something with AI” is the easiest sentence to say this week. The harder thing is starting the year with a system you can point to. Not a thread you saved. Not a demo you can’t reproduce. A real pipeline that ships.

So here’s our holiday gift for builders:

Enroll in AI Engineering by Jan 2, and we’ll gift you the 10-hour LLM Primer.

Two of our most loved courses, bundled for one outcome: from first principles to a production-ready LLM product.

👉 Claim the Gift Bundle

Most people building with AI get stuck in one of two traps. They either learn concepts but don’t ship, or they ship a quick demo without understanding why it works, then can’t improve it, evaluate it, or scale it.

This bundle solves for that!

The 10-hour LLM Primer gives you judgment. It’s the fast track to the mental models that stop you from guessing: when prompting is enough, when you actually need RAG, how to think about evaluation, and where agent workflows help versus where they create chaos. You finish with decision frameworks you can reuse across projects, not just notes you forget.

The AI Engineering course turns that judgment into a shipped build. This is where you implement end-to-end: you work through the real steps: designing the workflow, grounding the model with retrieval, iterating with evaluation, adding tools/agents where they improve reliability, and deploying a complete project you can demo. The goal isn’t “I watched lessons.” The goal is a production-style LLM product you can show, defend, and keep improving.

By the first week of January, most resolutions fade. The one that sticks is the one that produces proof. If you want to start 2026 as the person who doesn’t just talk about LLMs, but can ship them, a working product you can demo, defend, and keep improving, this is the cleanest path we’ve made.

This bundle is only live until Jan 2, 11:59 PM EST. After that, the Primer is no longer included as a gift.

A few quick details:

• What you get: AI Engineering + the 10-hour LLM Primer (free gift)

• When it ends: Jan 2, 11:59 PM EST

• What you walk away with: fundamentals + a shipped, production-style LLM product you can demo

👉 Get AI Engineering + Free Primer

Louis-François here, from Towards AI, the team behind “making AI accessible” for all of you.

This year, between our courses, book, newsletters, and other programs, we came closer than ever to that mission. People who were only “AI-curious” are now shipping tools, redesigning workflows, and talking about models with a real understanding.

To end the year, we want to push that mission a little further out into your circles, not just to you, but to someone you care about.

From December 11th to December 31st (11:59 PM EST), you can:

Gift any eligible Towards AI course and get your own seat in the same course for FREE.

One payment, two serious learners.

It takes about two minutes: pick the course, tick “this is a gift” at checkout, add their email, and we’ll unlock the same course for your account as well.

👉 Gift an AI Course + Join Free

For us, “accessible” doesn’t just mean “you watched a few videos” or “you tried a chatbot.”

It means:

• You can turn an idea into a working AI product, or

• You can use AI so deeply in your day-to-day work that your output and opportunities genuinely change.

Everything we teach sits on one of those two paths.

If your goal is to become an AI engineer, someone who can build their own AI product, start a small AI business, or design the systems other people rely on, you can choose our AI Engineering path. It has courses for every level, from “I’m new to Python” all the way to “I can ship a production-grade AI app end-to-end.”

If your goal is to stay in your domain but become the person who quietly rewrites how work gets done with AI, you can choose our AI for Work path. It’s built for professionals who want AI woven into their research, writing, analysis, decisions, and team workflows so they become the one people turn to when they ask, “How do we do this with AI?”

A few things to know (including how to claim your seat)

How to use the offer:

• Find the course that your friend needs

• Go to checkout

• Click the “this is a gift” option

• Enter their email and name

• Make your friend happy

• Enjoy the same course for free in your account within 24 hours

• Do the course together and learn

Quick details:

• Who can I gift?

  Anyone with an email address who’d benefit from learning AI seriously: a friend, colleague, team member, or family member.

• How does my free seat work?

  When you tick “this is a gift” at checkout and complete the purchase, we unlock the same course for your account as well. One gift = one matching seat for you.

• Can I gift more than one person?

  Yes. Each gift purchase gives you a matching free seat in the same course.

• What if we can’t start right away?

  You’ll both keep access and can follow the next available cohort or learn at your own pace (depending on the program).

• Are gifts refundable?

  No. All gift purchases are non-refundable, so please choose the course and recipient carefully.

• When is this offer live?

  The Christmas offer runs from December 11 to December 31, 11:59 PM EST. After that, gifting goes back to normal (no free matching seat).

🎁 Gift a course, take your own seat, and decide whether 2026 is the year you build with AI or the year you rebuild your work with it.

👉 Gift an AI Course + Join Free

Louis-François here, from Towards AI. If you were in the middle of life/work/BFCM chaos and missed our sale, I have good news: we’ve reopened our best deal of the year for you and tied it to the last AI cohort of the year.

We’ve extended our BFCM deal for one reason: to give you a final, low-friction chance to actually go from “AI enthusiast” to “AI engineer” in 2026.

TL;DR: Our complete From Coding Novice to Advanced LLM Developer path is now $599 → $359 (40% off). All courses are 40% off with the code “blackfriday”, but this is the journey we recommend if your goal is to enter the field as an AI engineer.

We’ve taken everything we know about getting someone into AI engineering and compressed it into one track:

From Coding Novice to Advanced LLM Developer

In practical terms, this bundle changes three things for you:

1. Code stops being the scary part.

You don’t just “learn Python”; you learn to code the way developers are now expected to work, with AI at your side. We show you the exact workflows we use to pair-program with models, build features, refactor, and debug faster. Instead of staring at errors and closing the tab, you become the person who speaks both languages: code and AI.

2. Your foundations are wired for production, not toy demos.

Across five intensive 2-hour sessions, you go from basic prompting to thinking like someone shipping real systems. You cover LLM fundamentals, RAG, evaluation, multi-step agent workflows, and large-scale optimization with one lens: would this survive in production? Your “basics” are aligned with how real teams design systems, not weekend experiments.

3. You ship an advanced LLM product.

In the final course, you build your first scalable product with LLMs, combining prompting, RAG, fine-tuning, and agents into a working application. By the end, you don’t just know the buzzwords; you have a real product you can open, demo, and defend in detail when someone asks, “So what have you actually built with AI?”

All of this is wrapped into one path:

• Bundle: From Coding Novice to Advanced LLM Developer

• Bundle price: $599 → $359 (save $240, 40% off)

• All courses: 40% off with code “blackfriday” (add at checkout)

• Format: Self-paced, lifetime access; designed to be completed in about a month of consistent work.

👉 Start your novice-to-advanced LLM developer journey for $359

(Code blackfriday, add at checkout. This is the best deal you’ll get on this bundle.)

This is Louis-François from Towards AI and the Gen AI 360 courses, and I want to show you what the next step looks like if you want to turn that foundation into real career outcomes.

If your 2026 goal is to move into AI engineering, lead AI projects, or push your compensation into the six-figure range, this BFCM is the cheapest that path will ever be.

With code blackfriday, our Full-Stack AI Engineering program drops from $349 to $209.

That’s 40% off and the lowest price we will ever offer for this course.

Full-Stack AI Engineering is where you go after Gen AI 360 if you want to:

• Go from “I understand how these models work.”

  to “I can design and ship an AI system end-to-end.”

• Build portfolio-ready projects you can actually walk through in an interview.

• Be the person who can make concrete technical decisions about LLMs, RAG, agents, and evaluation, not just talk about them.

If you start now, the next few weeks become build time.

January becomes the month you start applying for roles, updating your CV, or talking to your manager with real systems and demos to point to, not just “I’ve been experimenting with AI.”

All our courses are 40% off for BFCM, but this is the one that’s designed to get you to “full-stack AI engineer” as directly as possible.

If you don’t know Python yet, you can use the same 40% discount on Python for AI Engineering first, so you’re not blocked by code.

If you’d rather not code much but still want to be the go-to AI person in your team, you can use the discount on AI for Work and focus on using AI for writing, analysis, planning, and client work at a serious, professional level.

Same discount, different outcomes. You just pick the version of your 2026 you’re aiming for.

• All courses: 40% off with code blackfriday

• Full-Stack AI Engineering: Goes back up to $349 on Monday

👉 Explore the courses and claim 40% off

(Code blackfriday is auto-applied at checkout until Monday.)

Our November cohort of Full Stack AI Engineering closes in 2 days. The live kick-off call is this Sunday (November 2). After that, we start working with the people who joined. If you want to be in this cohort, enroll now and if you want to test it before paying, use the free preview below.

It’s the same curriculum we’ve used to train teams who paid $50,000+ for a structured, production-ready GenAI playbook. You don’t have to pay that. You don’t even have to pay today.

Enroll in the full course (join the November cohort): https://academy.towardsai.net/courses/beginner-to-advanced-llm-dev?ref=1f9b29

Current cohort pricing: $349 one-time

Or start with the free preview → check if the content, pace, and outcomes match where you want to go, and upgrade only if it’s the right move for your career.

🔗 Free preview: https://academy.towardsai.net/pages/free-lesson-offer-1?ref=1f9b29

🔗 Full course: https://academy.towardsai.net/courses/beginner-to-advanced-llm-dev?ref=1f9b29

Cohort starts Sunday (Nov 2) — sign up for the course to get access!

Why are we doing it this way? The AI engineer role is one of the most in-demand and most poorly defined roles right now. Everyone says they “do AI,” but very few can actually design a retrieval workflow, call models via APIs, choose between prompting, RAG, tools, and fine-tuning, and then ship something people can use. That gap is exactly what this cohort is closing.

Here’s what you unlock immediately in the free preview:

• You see the actual scope of an AI engineer, not just toy prompting, but the depth and breadth of the role, where engineering, product, and AI meet.

• You learn when to use Prompt Engineering, RAG, tools, and fine-tuning, and more importantly, why you can’t just “do RAG” for every use case.

• You get a walkthrough of calling LLMs via API because real work doesn’t happen in playgrounds.

• You get a complete syllabus breakdown so you know what you’ll build, in what order, and how it compounds.

And once you upgrade to the full course, you get access to the November cohort — a group of builders, professionals, and curious learners from around the world who are here to do one thing: ship real AI systems that solve real problems. Here’s what the full course is about:

I already reviewed early applications and chatted with students joining from startups, top universities, government, and companies like Nvidia, Deloitte, EY, and even police forces. This isn’t a passive “watch 200 videos” course; it’s a (virtual) room full of people doing the work.

The course’s goal is to teach highly practical industry-relevant skills through hands-onpractice, not by sitting and watching.

November cohorts are always more active because people want to ship something meaningful before year-end. If you want that energy, don’t wait for the next one.

It all kicks off with a live welcome call with our CEO on November 2, so if you want to join with momentum, enroll now — or preview and upgrade before Sunday.”

What to do now:

1. Enroll in the full course (to secure your seat)

2. If you want to test first, sign up for the free preview 👉 https://academy.towardsai.net/pages/free-lesson-offer-1?ref=1f9b29

3. Watch the intro lessons and the syllabus walkthrough.

4. If you can see yourself in this role, upgrade to the full course and join the November cohort.

No guesswork. No “I’ll see if this is worth it” in your head. You can see inside first.

See you there, and have a great weekend!

P.S. We have a 30-day guarantee refund if you don’t find the course is for you.

P.S.2. Reply to this email with any questions about this or our other courses, and I’ll be happy to help out!

Louis-François

If you’ve been waiting for the right time to start, this is it. Enroll in any course and you’ll automatically unlock access. Details will arrive in your welcome email (I’m also always available by email, just reply here :) )

One week from now, some learners will already be part of the October cohort — working on projects, comparing notes with peers, and building skills that directly translate to their jobs.

Others will still be scrolling, bookmarking tutorials, and waiting for the “right time.”

The choice is yours.

👉 Join the October cohort now — don’t put this off another month

What will you build this October?

By the end of the October cohort, you could:

• Code your first AI apps with Beginner Python for AI Engineering

• Master prompting, RAG, agents, tool use, and fine-tuning in the 10-Hour LLM Fundamentals

• Deploy advanced systems and portfolio projects in Full Stack AI Engineering (60 Hours)

• Or, if you’re not coding, design repeatable no-code workflows in AI for Work that make AI your daily collaborator

It all depends on the course you choose. Every path is hands-on. Every path is updated monthly. And every path starts with the October live kick-off on 5th October.

👉 Pick your course to join the October cohort

Why Join the October Cohort

Every Towards AI course is built for the industry:

• Monthly updates so you never fall behind

• Hands-on projects that push you to build and land jobs

• Dedicated course channels inside our 80k+ community

• A live kick-off every month: October’s is 1 day away

👉 Secure your spot before 5th October — the October cohort is almost full

One of the biggest challenges I see in our community is momentum. It’s easy to get excited about AI, try a few tools, and then stall when you’re unsure of the next step. Meanwhile, the field isn’t waiting—every week brings a new model, a new benchmark, a new workflow.

The question isn’t whether AI will shape your work, but whether you’ll have the skills to stay ahead of it.

And to help you keep moving, we’re offering a massive discount for the next 48 hours: 35% off all our courses.

You can start right away or join the October cohort when it begins.

Just add the code OCTCOHORT35 at checkout, only valid until FRIDAY!

You’ll find two pathways on the website, one for coders and one for non-coders. Pick the one that best fits you!

You’ll either walk away with

deployable MVPs and projects that open the door to six-figure AI engineering roles

or

repeatable system that makes AI a reliable collaborator in your job, not just a tool you test once in a while.

Platform Perks You Lock In

• Lifetime, self-paced access with ongoing updates.

• Certificates of completion for every course.

• 30-day money-back guarantee — zero risk to try.

• Optional access to our October cohort for live sessions and mentorship.

• And of course, the code OCTCOHORT35 gives you 35% off all courses — but only for the next 48 hours.

Find our courses here:

https://academy.towardsai.net/?ref=1f9b29

P.S. For the most motivated of you, the discount even works with our “Get it all” bundle, available for only $585 instead of $900!

I’m thrilled to finally share that we ( ) are launching the Towards AI Partner Program 🚀

Our goal with this big update is to allow anyone to earn by referring the course.

The idea is simple: help one person learn AI, and earn something concrete from the start!

1 referral → Our bestselling book Building LLMs for Production

3 referrals → Any one of our industry-focused courses

10 referrals → Full access to all our courses, and an invite to the affiliate tier (where you can start earning ~$70/course or $180+/bundle).

This isn’t about blasting links or spamming; it’s about helping out friends, colleagues, or your community to take their first real step into AI.

Just this year, our affiliates already earned $18,692.99 in 8 months, including $6,730.21 from just 3 posts.

We want to make this process accessible to all of you. So we added many assets, a special newsletter, a slack channel and more so you can share the courses more easily and get rewarded more easily.

For me, this program is a chance to make AI education accessible at scale while enabling others to build a reputation or even a side income. I'm always happy to refer products I love, and I think you will, too.

When I quit academia to work on AI full-time, I wanted to bridge the gap between research and real-world application. This program feels like a natural extension of that mission, empowering anyone to play a part in this, sharing the courses they love most, and getting rewarded for it.

If you’ve ever had someone ask you “Where do I start with AI?”, "How do I learn about agents?" or "what skills do I really need for my next job?" this is the answer. And now, helping them could change both their career path and yours! :D

👉 Read more and join here: https://academy.towardsai.net/pages/affiliate

Big thanks to the Towards AI team, and everyone who helped shape this program, and our current amazing affiliates, Paul Iusztin, Neil Leiser, Paolo Perrone, Smriti Mishra, Greg Coquillo, David Andres, Giuliano Liguori and all others.

GPT-5 just landed with big promises... but is there real progress?

Here's all you need to know about GPT-5...

By the way, we (Towards AI) are super excited to share that we are now creating DAILY short-form videos that cover real progress from AI updates in under a minute, including GPT-5, obviously.

If you want a quick, no-fluff "anti-hype" way to stay in the loop, you can now follow along on your platform of choice: Instagram, YouTube, or TikTok. :)

The latest few ones are all related to OpenAI, where I covered GPT-5, GPT-5-mini, GPT-5-nano and their two recent open models, oss-20b and oss-120b! I've attached the videos at the end of this iteration!

p.s. If you'd like to support my work, please do share these short videos with your friends!

Alright, let's get ot it.

It’s rolling out to everyone in ChatGPT, including the free tier, so you need to know about it.

There are three variants: GPT-5, GPT-5 mini, and GPT-5 nano.

What’s different when you use it? The model feels steadier. Fewer wild answers. Better code. Stronger reasoning. And yes, OpenAI leaned into the “PhD-level expert” vibe. Sam Altman said "chatting with GPT-5 is like exchanging with a PhD-level expert in all fields". But that's to be confirmed with time...

Under the hood. Context is larger: up to ~256k tokens today (they mention 400k even), so it holds long threads, docs, and repos in working memory.

There’s a “GPT-5-thinking” mode that lets it reason even longer when needed instead of burning cycles on every reply, and less on easier queries. This is because the model was optimized through reinforcement learning from human feedback (RLHF) (e.g. on real use cases) to spend more tokens when the task is complex, and less when the task is simpler. Thanks to this "think-on-demand", GPT-5 trims wasted reasoning, up to 50-80% fewer tokens on easy tasks (that they reported).

Like other recent models we covered in the short videos last week, it was trained to use tools directly, so it's a better "agentic" model than previous ones.

You can send text, images, and even video in. It reasons better about what it sees, then replies in text (or uses tools). It cannot generate images as with GPT-4o!

Coding is where the leap is clearest. On public benchmarks like SWE-Bench Verified, GPT-5 leads prior OpenAI models and edges recent rivals. It's even beating o3 and o4!

Knowledge cutoff is September 30, 2024, so you should be careful with that, but with live web browsing to bridge the gaps.

They've also optimized part of the RLHF training to reduce hallucinations (stick to facts), which is supposed to make it better than previous models, but it's still not perfect.

And most important of all is this sentence OpenAI shared: GPT‑5 is less effusively agreeable, uses fewer unnecessary emojis.

Finally!

Now, what do mini and nano change? Cost and speed.

Prices (API, per 1M tokens) : GPT-5 ≈ $1.25 in / $10 out. Mini ≈ $0.25 in / $2 out. Nano ≈ $0.05 in / $0.40 out. That’s ~5× cheaper to generate with mini, ~25× with nano, versus standard GPT-5. Latency drops as you go smaller.

So, which should you pick?

Use GPT-5 when quality beats cost. Deep reasoning. Multi-step agent flows. High-stakes coding or analysis. If the answer must be right and well-explained, this is the one.

Use GPT-5 mini for most day-to-day work. Everyday dev tasks. RAG and retrieval workflows. Product copy. Analyst notebooks. You keep most of the brains for a fraction of the price, and you’ll probably ship faster. It's the 80:20 of the model series.

Use GPT-5 nano for scale (when you want to spam the heck out of it!). Bulk summarization. Tagging and labeling. FAQ replies. Quick triage. When throughput matters more than nuance, nano pays for itself.

A few practical tips:

With GPT-5 thinking, let the agent work. Instead of “tell me how,” try “book the venue,” “draft the deck,” or “compare these three vendors and paste a table.” It will browse, log in when you approve, and hand you artifacts to edit.

In the API, adjust the new verbosity and thinking parameters to align with your goals!

There are two ways to tackle a new model suite release:

1. Test the best model. If it works, try the smaller one, and so on.

2. Test with the smallest; if it breaks, use a bigger one, etc.

Both are good approaches. Just don't use GPT-5 Thinking (most expensive one) for every type of query and task!

A quick recap: Start with mini as your new baseline. Reach for GPT-5 when the stakes rise. Keep nano in your pocket for bulk.

Don't forget to follow our content for similar daily updates with quick entertaining videos wherever you prefer: Instagram, YouTube, or TikTok!

Here are some examples as promised:

GPT-5 release:

oss-20b:

This email is your reminder to get to it and finally master LLMs!

Everyone’s trying them.

Few are making them work as they had hoped.

One day, ChatGPT or Claude speeds things up.

The next, you’re rewriting its entire output saying “it’s useless”. Most people end up wondering if it’s saving them time at all or just giving them more work in a shinier package.

The difference isn’t the tool. It’s knowing how to use it so it actually delivers.

That’s what AI for Business Professionals is designed to teach.

p.s. this course is for everyone, but we also have practical trainings for the most technical coders at the end of the email for this month’s cohort start!

What You’ll Actually Walk Away With

We’ve spent 3 years helping companies adopt AI training developers, guiding executives, and deploying systems that actually ship, and 3 more years creating educational content for everyone.

This new course distills everything we’ve learned into a system designed for busy professionals, managers, and developers who want to:

• Use AI with confidence, not just trial and error.

• Steer AI like the “CEO” of your assistant, getting results you can trust, not rewrites you dread.

• Know precisely where AI drives value in your role or company (and where it’s just hype).

• Treat AI like a capable junior teammate, leveraging its speed while catching mistakes and biases.

• Build prompting, context, and iteration habits that make AI outputs cleaner and faster than your own first drafts.

• Confidently use tools like ChatGPT Team, Claude, and Gemini without risking bad data practices or poor results.

Each lesson is short and actionable, so you can start applying what you learn immediately, even if you’re balancing a full workload.

“The course brilliantly cuts through the overwhelming flood of information on LLMs, delivering knowledge in digestible, bite-sized modules that build confidence step by step.” — Matt Chantry

Why Now

The full program is $399, built from years of work training teams and building AI systems across industries. We rarely discount it, but for this cohort, we’re offering 15% off, so more of you can join.

Use code august_cohort at checkout. The discount is valid until August 3rd, or until all seats are filled.

When you enroll, you’ll start learning immediately and automatically get a seat in our August 3rd live session with our CEO, where you can bring your toughest AI questions, pressure-test your ideas, and see how other professionals are using AI profitably, not experimentally.

Save 15% & Join the August 3rd Cohort →

Don’t spend another month stuck in trial-and-error while others make AI their advantage. Start now, build the skills to make AI second nature!

If you want to dive further into AI, check out our bundles and various trainings and find your best fit:

Still don’t know which course to take? Ask me and I’ll be happy to help!

Get started now!

I’m super excited to release our final 2-hour session of our "10-hour transition course" from developer to LLM developer!

Lesson 6 is LIVE: Fine-Tuning, LoRA, RLHF & Everything You Need to Re-train Your LLMs

If you’ve already nailed prompting and retrieval in the first two free sessions, you know that clever context takes you far, but not always far enough.

Tone, domain precision, stubborn hallucinations… sometimes you simply have to re-train the model. Especially when using a smaller (local) model.

What Lesson 6 delivers:

• When to Fine-tune an open (and closed) model

• SFT vs. LoRA vs. QLoRA—when each makes sense and how to avoid wasted compute

• PPO, DPO, GRPO, RLHF, RLAIF: a quick crash course in reinforcement learning for language models

• Failure-mode avoidance: over- vs. under-fitting, hallucination spikes, catastrophic forgetting

• Hands-on walkthrough with Unsloth—yes, even on free GPUs

“The course brilliantly cuts through the overwhelm, then hands you tools you can use the same day.” — Matt Chantry

Unlock the full course to watch Lesson 6 now: https://academy.towardsai.net/courses/llm-primer?ref=1f9b29