If you caught my post after Unleash last week — or any of my recent writing on this topic — you know I’ve been fascinated by what’s happening with in-house product building in HR. A small but growing cohort of TA and HR leaders have quietly crossed a line, moving from tinkering with niche AI workflows to vibe coding real, sophisticated products that are beginning to replace expensive vendor contracts. Compensation intelligence tools. Engagement survey platforms. Application review automation. Built internally, by practitioners, with no engineering team in sight.
The response to that post made one thing clear: this isn’t an isolated observation. It’s a signal worth taking seriously. So I want to go deeper.
The ‘Access’ Wall is Entirely Gone
Build vs. buy has always been a conversation in this industry. But “build” was mostly aspirational – too slow, too costly, too dependent on engineering resources that HR and TA teams never truly had meaningful access to. That constraint is now collapsing fast.
What once required a technical co-founder and a startup budget can now be done by a motivated practitioner with Claude, Zapier, Vercel, Replit, or any number of vibe coding platforms available today. I lived an earlier version of this myself building TalentWall – but what took years and a company to pull off then, a Head of Talent can essentially do on a weekend now. The barrier to entry hasn’t just lowered. For a specific category of internal tooling, it’s gone.
And this isn’t isolated builders working alone in the dark. A community is forming. Ashby recently hosted an event where recruiters showed each other live what they’ve built – hiring playbooks generated in minutes instead of hours, self-serve ATS dashboards, candidate-facing job packs deployed via URL. At Unleash, I found myself standing outside a happy hour with a handful of practitioners, everyone pulling up the apps they’d built like kids showing off what they got for Christmas. The energy was different from anything I’ve felt at a conference in a long time. The practitioners who’ve crossed the 0-to-1 threshold aren’t looking back, and more are crossing every week.
Not Everything Should Be Built
To be clear, this isn’t a “build everything” argument.
The calculus that’s changed applies to a specific category of tools: internal analytics, workflow automation, manager-facing dashboards, compensation benchmarking for internal use, engagement intelligence. Low-stakes, non-candidate-facing, recoverable if something breaks. These are the tools where the vendor lock-in is weakest, and the buyer/user frustration has historically been highest.
There’s a separate category where you should keep buying: regulated functions like payroll, benefits, equity management, anything with direct legal exposure on candidate decisions. One person in my comments put it well – mess up an equity edge case and someone could lose a million dollars. The vendor accountability transfer still has real value there, and “we built it ourselves” is not a satisfying answer when litigation comes knocking.
The line isn’t always clean, and it will continue to move as AI advances. But knowing roughly where it is matters before you start.
It’s Already Happening
Literally a couple hours after my Unleash post went up, I came across a job posting from a former colleague at Thumbtack: Senior AI Solutions Builder, People. The role sits within the HRBP team – not IT, not Engineering, not Product. It requires a specific and new kind of profile: part product builder, part HR strategist, part AI practitioner. The JD explicitly calls for evaluating build vs. buy tradeoffs. A likely first project? Building an AI-powered tool that gives managers on-demand access to HRBP-caliber coaching – replacing a category of vendor interaction with an internally-built system.
Thumbtack isn’t a fringe experiment. They’re a scaled, well-funded company making a deliberate strategic investment in internal AI building capability embedded directly in their People function. When companies start creating dedicated headcount for this, it stops being a skunkworks trend and starts being strategy.
Implications for Vendors
This pattern isn’t entirely new. Large tech companies have done a version of this for decades – using platforms like Workday and Oracle primarily as administrative infrastructure while building everything strategic on top (thanks Steve Hunt for these insights!). What’s different now is that AI is making that path viable for companies that never had the engineering capacity before. The behavior that was previously limited to a small club of well-resourced tech employers is becoming accessible to almost everyone.
So the question is: do we eventually see platforms where companies pay for the secure foundational layer – data infrastructure, compliance guardrails, system of record – while the UI and workflow layer increasingly gets built internally? It’s not guaranteed. Plenty of companies won’t have the appetite or talent for it. But for the ones that do, the value equation with their vendors is already shifting.
The vendors I’d be watching most closely are the ones whose products have historically generated the most buyer frustration – the tools that covered 60-80% of what teams actually needed but never quite all of it. Think most ATS platforms(sorry not sorry), those valuable but rarely fully adopted Talent Intelligence platforms, as examples.
That gap is exactly where internal builders are starting. They’re building zapier and n8n workflows that, in an ideal world, should already be part of the underlying solutions. The smarter vendors will recognize this as a signal to evolve how they build and how they charge. The ones who don’t may find themselves getting quietly hollowed out from the edges before they see it coming.
Five Things I’m Watching
Predicting anything beyond 6 months as it relates to AI is and probably always will be a fools errand. But here’s where I’m paying attention, and I’m genuinely curious whether others are seeing the same things:
Internal builder roles proliferate. Roles like the Thumbtack one feel like they could become standard headcount on People teams at scaled companies – but how fast that happens, and whether it sticks, probably depends on how much value the first wave of hires demonstrably delivers.
Vendors experiment with builder-friendly configurations. I’d expect at least one major HR software vendor to launch some version of an API-first or headless offering targeting internal builders. Whether that becomes a real business model or a footnote is harder to call, but my gut says it’s the former.
Low-frustration tools hold, high-frustration tools don’t. The products most at risk are the ones where buyers were already unhappy – analytics layers, reporting tools, manager-facing workflows. The ones with real data gravity and regulatory moats – payroll, HRIS core, equity – feel safer, at least for now.
A credentialing layer emerges for builder skills. Portfolio-based, practitioner-validated proof of what someone has shipped – not certifications for knowing how tools work, but evidence of building something real. I think this becomes a meaningful hiring signal, though I’m less sure of the timeline or what form it takes.
The skills gap compounds. The practitioners building now are accumulating an advantage that’s hard to close later. Whether that leads to a new kind of career stratification in HR and TA, or whether the tools get accessible enough that the gap closes on its own, is one of the more interesting questions I’m sitting with. Probably a future post coming on this topic on its own.
As always, it’s still early innings. A lot to watch. But the hallway conversations don’t lie, and this one’s been getting louder.