Jason Miller has a name for this: selfish software. You build it for an audience of one. The stuff that used to die in your head because it wasn’t worth a real project. It’s worth it now.

Build 1: A sourcing system that remembers my taste

I have around 8,679 LinkedIn connections in a queryable database. Every role gets a spec, in my own words, about who actually fits. Then I review the people it surfaces, keep or cut with a note on each, and those decisions flow right back into the spec. Next time I run that role, it’s sharper. My taste used to reset to zero after every search. Now it compounds.

This one’s a whole project on its own, and I’ll go deep on it in its own post soon. Today it’s just the setup, because it’s the reason I needed the second thing.

Build 2: A morning briefing agent, running on the NAS in my closet

My laptop closes, sleeps, comes with me. A scheduled job is only as reliable as whether the lid is open. Then it hit me. My NAS is always on. It just sits there holding files. So I asked Claude the dumb question, and the answer was yes.

So I built this. Every morning at 7am, before I’m up, a job on the NAS reads the last 24 hours of my Gmail, hands it all to Claude, and Claude writes me a briefing. Not a digest of what people said. A list of what I need to DO. It groups everything into Urgent (respond today), Important (this week), and FYI, then gives me a numbered to-do list, most important first. It lands in my inbox as a clean formatted email by the time I’ve got coffee in hand. I pick up my phone and I already know my day.

The whole point of a 7am briefing is that it has to run without me. If it depended on my laptop being open, it wouldn’t be a briefing, it’d be a chore I forget. On the NAS it just happens. Always on, no cloud bill, nothing leaving my house.

And I want to be honest about something. I’m a recruiter. I can’t really write code. I built all of it anyway, with Claude. It all starts with an idea turned into a prompt. If you can imagine it and reason through it, you can build it.

Signing Off

Both of these are selfish. Nobody asked for them. And both went from idea to production in a single evening. The cost of building for an audience of one has fallen through the floor. And if you’re just getting started, the first step is the first step.

Not a solution. Not a feature list. A clear answer to: what problem are we actually solving, and for whom?

That stuck with me. Especially now, when everyone I know is spinning up AI projects and automations. Some are genuinely useful. A lot are cool demos that solve a problem nobody really has.

The difference almost always comes down to whether someone scoped the problem before they started building.

Two questions I ask before starting anything

1. Is there a real problem, who feels it, how often, and what are they doing about it today?
   
   Not “everyone has this problem.” One person, one situation, one real workaround. The workaround tells you more than the complaint.
   

2. Do you actually have what you need to build it?

   
   The data, the access, the integrations. If you’re assuming you can get it without confirming, that’s a risk worth surfacing before you write a line of code.

Hand-waving through either of these is a sign to pause.

What a good problem statement looks like

Bad: “Recruiters waste time on repetitive tasks.”

Better: “Sourcers at early-stage startups spend 3+ hours a week manually formatting candidate summaries for hiring managers, leading to delayed feedback and slower time-to-fill.”

One tells you nothing. The other tells you who to build for, what it costs them, and what success looks like.

The skill I built to help with this

I turned this framework into a Claude skill called to-build-or-not-to-build. It’s a guided interview that asks you these questions one at a time, probes weak answers, and ends with a clear verdict: Build it, Don’t build it, or Validate first. Then it outputs a written problem statement you can actually use.

It’s open source at github.com/talent-collective/to-build-or-not-to-build. Grab it, fork it, make it better.

AI founders are fighting each other (and Claude) for survival right now. The competition is brutal, feature velocity is relentless, and the pressure to stay ahead is constant. I have a lot of empathy for that position. Genuinely.

But that pressure doesn’t stay at the top. It rolls downhill. And what I’m seeing more and more is founders expecting their teams to absorb it the same way they do.

That’s the problem.

A founder’s relationship to their company is existential. It’s their name, their equity, their identity. If it works, the outcome is genuinely life-changing. If it doesn’t, they carry that.

An IC’s relationship is different. Not worse. Different. They care. They show up. They believe in what they’re building. But there’s a ceiling on how invested they can be, and that ceiling is set by how much skin they have in the game.

If they had the same stake as the founder, they’d be a founder.

There’s a financial piece here that doesn’t get talked about enough. $300K sounds great on paper. Divide it by 60 hours a week instead of 50 and the effective rate looks very different. The expectation to grind has to be proportionate to the upside.

If you feel this way right now, still sitting at your desk, still doing the work, but running on fumes, you’re not mistaken and you’re not lacking commitment.

You’re being asked to fund someone’s dream with your burnout.

From first prompt to a live, publicly accessible tool with automated weekly updates took just under two hours. Here is exactly how I did it.

Subscribe now

Problem

I regularly find myself hunting through the same handful of PDFs: Ashby’s benchmarks, Gem’s outreach data, LinkedIn’s talent report. I know the data exists. I just can never remember which report has which number, and copy-pasting context into Claude every time I want to ask a question stopped being sustainable.

I wanted one place where all of it lived, queryable in plain English, always up to date, shareable with anyone in TA.

Build Process

Step 1: Collect the reports

I started by identifying the reports I actually reference. Ashby, Gem, LinkedIn, SHRM, iCIMS, Korn Ferry, HireVue. Then I prompted Claude to go get them:

“Look up all reputable 2025 and 2026 recruiting trend reports.”

Claude pulled each report, extracted the meaningful data, and saved everything as structured markdown files. For each one it captured the methodology, sample size, key stats, and direct quotes so they could be cited properly later. I ended up with 22 reports.

Step 2: Make them queryable

A folder of markdown files is not useful to anyone but me. I wanted something anyone could connect to Claude and use in plain English. That meant building an MCP server. MCP (Model Context Protocol) is how you connect external data sources directly into Claude so it can reference them in any conversation, without you having to paste anything in.

My prompt:

“I want to create a function that allows anyone to reference this folder of markdown files. It could be a Claude skill or an MCP.

From there I layered in specific capabilities one prompt at a time such as:

“I also want a feature to search for quotes across reports.”

“When users query the MCP, it’s important that you ask them what year they’re looking to query. Some people may want to compare year over year.”

Step 3: Deploy it publicly

I told Claude I wanted it publicly accessible. It walked me through pushing to GitHub and deploying on Render. The server has been live since.

Step 4: Automate the updates

“I want to create an agent that scours the web weekly for new reports, reads them, and adds only credible reports to my knowledge base. Send me an email when a new report is added.”

A scheduled agent now runs every week, evaluates new reports for credibility and methodology, and adds qualifying ones automatically. I do not have to touch it.

Solution

The result is a public MCP server with 22+ recruiting benchmark reports. Anyone with a Claude account can connect it in 30 seconds:

1. Settings > Customize > Connectors > Add custom connector

2. Paste in: `https://two026-recruiting-reports.onrender.com/mcp`

Then ask it anything:

• “What’s the average time-to-fill for a 500-person company?”

• “What does the data say about offer acceptance rates at Series B startups?”

• “Find quotes on AI in hiring from 2026 reports”

Full source and report list: github.com/talent-collective/recruiting-reports-mcp

---

Signing Off

The immediate use case is faster answers to benchmark questions. Instead of “I think I saw something in the Ashby report,” you get the actual number with the source and sample size in about 10 seconds.

The larger bet is that this is how TA teams should be using AI: not just as a writing tool, but as a live connection to the data and research that informs decisions. Most teams are still copying and pasting. Building an MCP means the context is always there.

All you need is a goal, a decent sense of how you might approach it, and Claude to help you get there. The prompts are the product. The code is just what happens in between.

Your turn.

For those who don’t know, dev.scout surfaces developers and helps recruiters evaluate talent through their open source contributions, not their resume. I built it. I’m a recruiter. I can’t write “hello world” . I built it anyway.

But here’s the thing: I didn’t just want it fixed. V1 was frankenstein’d together over 200+ hours across a month. I’d learned as I went, hit walls, backtracked, duct-taped things that should have been built properly. I knew the models had gotten significantly better since I first built it. So I handed Claude Code the whole codebase and basically said: rebuild this the right way.

Five hours later it was back up and deployed. Better than before: more resilient, less buggy, and most importantly, maintainable and extensible.

The thing that got me wasn’t just the speed. It was what happened after. With v1, every time I tried to add something or fix something, something else would break. Pure whack-a-mole. This time I could make changes and they’d just... work. That’s when it actually hit me.

This is the part where I’d usually try to explain it but I’m not sure I can. Claude Code read the entire codebase, understood what it was supposed to do, and rewrote it. 55 files touched, 4,257 lines changed, every API route, every service, middleware, config. All of it. In an evening.

This felt different. Not just faster. Like a real shift in what’s actually possible.

dev.scout is completely free for the month of March. Please try it and lmk what you think!

Now what to build next…

I ask everyone two questions. They sound simple. They’re not.

“Tell me about a role you know really well.”

What I’m listening for isn’t a role summary. I can read the JD. What I want to know is: did you go beyond it? Do you know what makes this role different at your company vs other companies? Do you know the team dynamics or failure modes of the last hire?

If I’m a candidate getting on a call with you, I expect to walk away knowing something I couldn’t have Googled. The best recruiters are operators. They’ve done the homework. They can tell you why the last two people in this role didn’t work out, what the hiring manager actually cares about versus what’s written down, and where you’d be set up to win. They take it personal.

What I got instead was a lot of role recitation. Almost verbatim JDs. That tells me the recruiter is a messenger, not a locked-in partner.

“Tell me about a candidate you lost.”

This one’s really about five things: growth mindset, storytelling, quarterbacking the close, adaptability, and deliberate urgency.

One pattern I kept seeing: surface-level answers. “They took another offer.” Okay, and? What did you learn? What would you do differently?

Almost every time, I had to ask that follow-up myself. And the thing is, if you’re the type of person I want to hire, you’ve already asked yourself that question a hundred times. You offer it up before I can even get there. The introspection isn’t something I pull out of you. It’s already sitting at the top.

The best closers don’t just work their own lane. They quarterback. They make sure every person on the field knows their role and executes. The hiring manager, the coordinator, the interviewer panel. Everyone is aligned, everyone is moving with purpose, and nothing falls through the cracks because the recruiter is driving it.

And when the play breaks down, they adapt. They don’t run the same route when the coverage has changed. The best recruiters I’ve worked with know when to go off script. They read the situation and adjust. The standard playbook is a starting point, not a ceiling.

Urgency is the thread that runs through all of it. Not panic, not pressure. Deliberate urgency. The kind that comes from genuinely caring about the outcome.

These aren’t the only things that make a great recruiter. There’s plenty more. But in my experience, these are the behaviors that separate good from the best. The gap lives in the details most people don’t think to care about.

And I want to be honest about something. This is a brutal market for recruiters right now. A lot of talented people are grinding through a search that’s taking way longer than it should. I feel for them genuinely. I hold this bar because I respect the craft, not because I’m dismissive of the struggle.

I hope this is helpful for all those out there on the grind looking for their next home. I know I’m not the only TA leader who holds these behaviors in high regard.

Subscribe now

Last week I wrote about Talent Engineering and Operations. This week I’m hiring for a role that on the surface looks like every other technical recruiting role: Lead / Staff Technical Recruiter.

It is, kind of. The remit is the same: own end-to-end pipeline for Engineering, Product, and Design. Partner with hiring managers. Source, screen, close.

What makes it different isn’t the job description. It’s the fit.

The thing nobody says out loud about recruiting roles:

People pick companies based on money, brand, and reputation. Recruiters are no exception — and that’s the irony. We spend our days coaching candidates to look beyond the name, dig into culture, do their diligence. Then we turn around and take the biggest offer from the most recognizable logo. We rarely take our own advice.

Not often enough do people invest the time to figure out whether it’s actually the right fit. And honestly, even when they do, it’s not easy to tell from the outside. You’re not going to see the skeletons in the closet from the front door. You find out once you’re inside — hopefully on a tour, not moving in.

Every team has its jagged edges. The question as a candidate is whether those jagged edges line up with yours.

So I’m going to tell you exactly who thrives here. Not to filter people out. To save everyone’s time.

The wiring that matter:

These can’t be learned on the job. They’re either part of your DNA or they’re not. If you’re missing even one, it’s probably not the right fit.

1. Urgency isn’t a behavior, it’s a natural state. The best candidates are off the market in days. You don’t wait for things to happen; you make them happen. Urgency is just how you’re wired.

2. Unapologetically intense. You operate with an intensity most people don’t. When a detail gets missed, you take it personally. That’s not a bug — it’s a feature.

3. Permission is not required. When a channel isn’t working, you build a new one. When a process doesn’t exist, you create it. You don’t wait for someone to hand you a playbook — you write it. High agency isn’t something you turn on for big moments; it’s your default setting.

4. The 2010-2020 structured interviewing playbook is no longer gospel. The landscape has completely changed. If your only play is creating structured interviews and running a consistent process, this role isn’t for you. You design interviews that meet the moment and allow us to win the best talent.

5. You’re one of them. You’ve spent enough time in the engineering world that you’ve developed genuine technical fluency. When you’re in a room with engineers, you belong there. That credibility is what lets you assess them honestly and earn their respect in the process.

Full JD here if you want to go deeper.

If your jagged edges line up with ours, let's get married (iykyk). DM me.

Last year, I wrote down what I called “Talent Engineering” - the idea that recruiting should be rebuilt like a product, not managed like a mess of duct-taped tools.

The thesis was simple: Modern LLMs, MCPs, and open APIs mean we can finally build recruiting systems that don’t suck. Instead of wrestling with ATS + CRM + Slack + Airtable + Zapier + Notion, we can ship intelligent workflows that make recruiters faster and hiring feel magical.

I knew it was inevitable. But when I came into this role, there was so much foundational work to do first - interview processes to build, pipelines to get flowing, basic systems to stand up. In an ideal world, I’d do both: lay the foundation and ship the AI tooling. But as one person, that’s just not possible.

And the hiring needs can’t wait. We need those sourcing bots generating pipeline now, not after I’ve finished building everything else.

Three things converged this weekend that made it clear: it’s time to hire for this.

1. Diversifying our Top of Funnel

Coming into AirOps, we’ve relied heavily on agency partners and referrals, with some outbound and inbound mixed in. To scale our hiring, we need every channel firing on all cylinders.

The foundation is working: my early sourcing has been well-received, the team collaborates on referrals, and our agency partners are crushing it. Now we need someone who can build on that foundation and turn our underutilized channels into reliable pipeline generators.

2. Ben, our new GTM Engineer, showed me this over the weekend:

Seeing what he could build in his spare time confirmed what I already knew: the right builder with recruiting context is be absolutely goated.

3. I finally accepted that I’ll never have the bandwidth to be that builder

As Head of TA, my time goes to hiring managers, pipeline reviews, and fire drills. The strategic projects that could 10x our recruiting efficiency? They’ll never get built if I’m the one building them.

So we’re hiring a Talent Engineering and Operations Lead.

This person will own the projects I’ve been dreaming about:

• Sourcing bots that generate qualified pipeline at scale

• A market intelligence system that tracks competitor moves and candidate activity

• An internal referral and backchannel database that captures institutional knowledge

• AI-powered candidate prep and personalization that doesn’t feel robotic

You’re the right person if:

You have the technical chops

• You can whiteboard how you’d implement each project above

• You started vibe coding a solution halfway through this post

• You’ve built point solutions before and swore you’d never make that mistake again

You understand recruiting operations

• You’ve spent 1+ years as a full-time sourcer or recruiter

• You’ve felt the pain of wrestling with broken systems daily

• You know what good candidate experience actually requires

You think like a product builder

• You approach problems with the user at the center

• You ship small, iterate fast, and measure what matters

• You get excited about building leverage through systems, not headcount

Why AirOps is the place to build this:

We build products for growth teams. Creative growth solutions are literally in our DNA. You won’t spend six months convincing leadership that experimental recruiting systems make sense. You’ll be surrounded by people who already think this way.

Plus, you’ll have Ben (our GTM Engineer) to collaborate with on the recruiting use case.

If this sounds like the job you’ve been thinking about since GPT-4o, DM me. Let’s make it a reality.

I thought this would be a good time to reflect and share what I’ve been up to since beginning of Jan.

What my week looked like:

// gnarly cough and congestion to start the week 😷
// 18 Recruiter Screens
// cleaning up pipelines with 1000+ candidates (thanks Covey)
// debugging out VPE pipeline
// taking on 3 new roles that I don't know anything about
// trialing a new sourcing product
// onboarding a virtual/off-shore RC
// 2-hrs of leadership meetings
// my first professional coaching session -- why did I wait so long?
// supporting my team as they develop relationships w/ their stakeholders
// brainstorm session about employer brand sourcing session
// HC review: what's open, closed, delayed, reconciling target start dates
// putting together weekly reporting to share with the business
// be semi-present as a husband and father

Needless to say, WE ARE SO BACK.

Three Realizations

Don’t try to change anything; get some wins first.

1. Most companies are operating just fine even if it’s not optimal. It may even seem insane as someone coming in — it certainly did in my case. Fact is, they were able to get to this point operating the way they are. Best thing you can do is put your own opinions aside and join in the chaos.
   
   Imagine joining a pit crew in the middle of a F1 race and you try changing how they are operating in the moment. You are surely to do more damage than be helpful. When the current race is over, you’ll have your moment to make the change you think are most critical. Be a good teammate and help the team finish that race first and earn your right to have an opinion.
   
   Every book will tell you to do this but I don’t feel like it emphasizes fighting your human instincts to want to change things to your version of normal. It’s counter-intuitive making it especially difficult. Anyway, I hope my analogy helped.

1. Head of TA is not a glamorous job

   
   As you can see from my list of things from this week, none of those are fancy by any stretch. As a Head of TA, I’m doing the same job as my team and then some. There is no room for strategy work at this stage. There are so many foundational pieces that need to be put in place before the ‘cool’ work can begin. There are countless things I would love to automate, apps I want to vibe code, sourcing bots I want to configure. There just isn't time right now—not until the foundational pieces exist.

   
   All that to say, I really need to find the mental space to zoom out and build a proper roadmap for the team beyond hiring. Even if it’s one small step per week, we need to start making more progress on the broader TA function at AirOps.

   
   

2. Who you work with makes all the difference

   
   Work is fun again! As intense as this week has been, I can honestly say it’s been so much fun (excluding being sick). I attribute this to really enjoying the people I work with everyday, legitimately believing in the product we’re building/selling, and finding micro wins throughout the day. When I reflect on my time at Vercel, I had maybe 2/3 of these; at Netflix, it was also 2/3. At AirOps, it’s fasho 3/3.
   
   If I could describe the people at AirOps with one word, it would be helpful. I took over a pipeline that was being managed by 3 different leads on the team. I needed to figure out the status of 30 different candidates. I shared a spreadsheet out to a Slack channel for them to fill out. It was 80% filled out in 30-minutes. THIRTY MINUTES! I’ve asked for things less involved than that at other companies that takes days. And it’s not like they’re not busy; it’s the fact that they prioritize helping those around them. This is just one of many things i’ve experienced since being here — the vibes are so good.
   
   I won’t go into our product beyond saying — AirOps is inevitable.
   

   In terms of micro wins, the best one came from one my stakeholders.

   

   I've always prided myself on fully embedding myself on the teams I support. This may not seem like much of a win but to me, it SOOO reinforcing.

A little more than a month in and I absolutely 🔒 in. Hoping I have more to flex about the next time I write.

Btw, we’re hiring (had to do it)

I’m a father of two young boys. That means I need to be selective about where I invest my time.

Over November and December, I spent time with 7 VC Talent Partners and 36 founders as I thought carefully about what I wanted to do next. I’m deeply grateful to the friends and peers who were generous with their time, advice, and encouragement throughout that process.

I wasn’t just looking for a new role. I wanted to be confident that I was building with the right people and doing the best work of my career.

Here’s how I built conviction in that decision.

The Founders

Matt, our Co-founder and COO, was my recruiter throughout the process. What stood out immediately was how collaborative, curious, and respectful he was. In our conversations, it felt like he was genuinely trying to learn from me rather than grilling me to suss out whether I knew what I was talking about. There was a humility in how he showed up that I don’t see often.

That humility made me curious. I’ve never worked for a winning founder or leader who wasn’t unreasonable in some way. They all have their own idiosyncrasies. That “unreasonableness” is usually just conviction showing up clearly somewhere.

So I put my research hat on and talked to other execs, asking the questions I always ask in situations like this: what are the founders relentless about, where do they push hardest, and what are they annoying about?

The answers were consistent. Their relentlessness shows up in customer obsession. They push hardest on prioritization and focus. And if there’s anything they’re “annoying” about, it’s taste. Their intensity doesn’t show up as frenetic or performative energy. It shows up in being opinionated about the things that matter and immovable once they’ve decided.

Recruiting isn’t Matt or Alex’s deepest area of expertise, so it made sense that they approached our conversations with curiosity. What mattered more was seeing that the same humility was paired with real intensity in the parts of the business they know best.

Collaboration and curiosity aren’t a counterweight to winning here. They’re the mechanism.

I don’t think there’s such a thing as a founder who isn’t some flavor of crazy. As you think about your next move, the work is figuring out whether their kind of crazy matches yours.

Product Culture

I didn’t know much about SEO, let alone AEO, before engaging with AirOps. What changed wasn’t a single conversation or deck, but a pattern that kept repeating.

Customers spoke about AirOps with real specificity. It wasn’t abstract praise; it was clear how the product showed up in their day-to-day work and why it mattered. That kind of signal is hard to fake.

The same traits I saw in the founders showed up in how the product was built and how customers were treated. There’s a tight feedback loop between users and builders, and a strong bias toward focus and quality.

AirOps reminded me of Gem in the early days. Customers were obsessed with the product, and we were obsessed with our customers. That dynamic creates momentum when it’s real.

What also stood out: AirOps invests in the broader growth and content marketing community through education and certifications. That said a lot to me about how they think about their role in the ecosystem and not just extracting value but building it.

Reimagining Recruiting

I wanted my next role to give me space to rethink recruiting from first principles. The playbooks that worked in the early 2020s don’t map cleanly to the AI era, and I wanted to be somewhere that recognized that.

AirOps is that environment. The leadership team values prioritization, understands the importance of slowing down to speed up, and recognizes that recruiting teams need real investment to succeed.

One moment reinforced this. During a casual lunch with part of the leadership team, we started brainstorming how our own product could be used to generate thought leadership content and, over time, drive organic traffic to open roles. It wasn’t about tactics; it was about leverage.

The conversation lasted maybe five minutes, but it told me everything I needed to know about how this team thinks about experimentation and building systems rather than just filling seats.

Let’s Connect

Over the course of 2026, I’ll be growing the recruiting team at AirOps. I’m looking for recruiting polyglots: former Heads of Talent and first recruiters with low egos, strong judgment, high agency, and a genuine desire to build a world-class recruiting function.

I’m not looking for direct reports. I’m looking for partners.

If what I’ve shared here resonates, please reach out. I’d love to connect.

Flowers 🌸

I want to call out a few people I’m lucky to call friends who directly or indirectly helped me during this search: Allison, Amina, Brie, Caro, Chris, Erin, Glen, Jason, Katie, Lauren, Mario, Nolan, Peter, Tammy, Sierra

Here’s my story and reflections of going from TA to Recruiting Operations.

Coming out of One Medical in 2020 after the IPO, I took a real pause to figure out if I was ready to run the same Head of Talent playbook again after a very intense sprint. Pretty quickly, I realized I was not, so I did what a lot of us TA leaders do. I stepped into consulting.

I still remember talking with Steve while I was debating my next move. He said, “If you are planning to consult, why not work with Gem?” I already loved that team, so it was an easy decision. What started as a project focused on internal processes and a few key hires eventually grew into leading the Customer Talent Advisory function for almost three years.

During that time, I partnered closely with recruiting and operations leaders on systems, tooling, analytics, and broader recruiting strategy. It was my first deep dive into operations.

That drive to understand operations end to end led me to Netflix, where I managed programs across referrals, structured interviewing, internal mobility, and the Eightfold+Workday implementation. I learned a lot about scale, culture, and what strong program management looks like at a global company. And while I loved my team at Netflix and met incredible people, I missed the pace and feedback cycles you get at a startup.

I connected with Amina shortly after meeting Amanda at RecOpsCon. The talent on the People team was undeniable, and paired with my itch to be back at a startup, joining Vercel was an easy decision. Vercel gave me a very different type of operational experience. The company was moving fast and hiring aggressively, and because velocity was the priority, RecOps functioned more as a support engine than a strategic driver. My team still owned systems, tools, workflows, headcount planning, reporting, and AI transformation, but the mandate was clear: keep hiring moving, even when typical ops tenets like consistency and structure had to take a back seat.

More than a dozen recruiters have reached out asking how I made this transition. I share this story, but I also share my honest perspective. Operations can be fun and incredibly high impact, but only when the company, the stage, and the leadership team truly value it.

What most people do not recognize and appreciate is the amount of change management involved. A huge part of RecOps is convincing recruiters to do things differently even when it slows them down, does not help them hit their numbers faster, and solves for something they are not incentivized to care about, like compliance, data hygiene, consistency, or risk mitigation. It often means asking people to absorb short term friction for long term health. It is meaningful work if the company is at the right stage to value it.

If you are considering making a career change into Recruiting Operations, evaluate whether RecOps is positioned to be a strategic lever or a necessarily evil as a means to an end.

I genuinely enjoyed this tour of duty, but it also clarified something important. I missed being on the front lines, partnering with functional leaders, building teams, and scaling companies directly.

So I am coming back to my roots.

I accepted an offer to join an exciting AI startup as their Head of Talent. It feels like the right moment to take everything I learned across Gem, the customers I worked with, Netflix, and Vercel, and bring it back to the work that energizes me most, which is building talent dense teams that help companies grow.

More to come on my next play.