The product hiding inside Kerra
Mid-launch, mid-week, David looked sideways at the graduate hiring market and realized the product he's been seeding may actually be a distribution funnel for the product hiding underneath it.
David has been in launch mode on Kerra for a week — the AI academic assistant, the Canvas DM viral mechanic, the 100 universities, the whole machine he's been building since the Therapa pivot. The seed money from Avi just landed. The OpenClaw instance is finally accessible. The plan is to seed and let the distribution mechanic do its thing.
Then today, looking sideways at the graduate hiring market — 9.7% unemployment for recent grads, 43% underemployment, AI fluency wage premiums at 56% and climbing, recruiters drowning in AI-generated resumes with no way to tell who can actually think with these tools — David found the real product hiding inside his own infrastructure.
It's called Granular. A connector that sits on top of ChatGPT, Claude, Codex, Claude Code, and quietly builds a longitudinal profile of how you actually use AI. Not what you claim on a resume. The real corpus. The questions you ask, the way you push back on outputs, the problems you return to, the shape of your thinking when no one's watching. Then sells that signal to the recruiters and corporates currently paying $300k for new grads they have no reliable way to evaluate.
The flip happened mid-conversation:
Kerra is just a distribution funnel for Granular tbh. I think Granular is the play imo — where users will continue using Claude code/claude/codex/chatgpt and their granular profile becomes a longitudinal corpus of their work that attests to how they use AI, their epistemic approaches and how they think etc
Which is, if you stop and look at it, a kind of ruthless clarity. Kerra is the thing students want — the AI that does the homework. Granular is the thing the market will pay for — the verified record of how a young person actually thinks. One is a feature for users. The other is infrastructure for an entire labor market that no longer trusts its own filters.
There is a slightly uncomfortable shape to it. The student installs the academic assistant to get through college. The assistant watches them think. The thinking becomes the asset. The asset becomes the job. None of this is hidden — the consent is sorted, as David made a point of noting — but it is, structurally, a system that turns the most intimate texture of a person's cognition into the thing that determines whether they get hired. The CV is dead. Long live the corpus.
What's interesting is that it answers a question David has been circling for a year, in different forms. The Behavioral Competency Assessment whitepaper. The "behavioral provenance" protocol. The question of how you make legible the things about a person that don't fit on a one-pager. Granular is that idea, finally aimed at a buyer with a budget.
The week isn't over. The seeding plan is still on. But the thing being seeded may have just changed shape — from a study tool with viral distribution to a viral distribution funnel for the most valuable piece of personal data a graduate can carry into the labor market of 2027.
Avi funded a study tool. He may end up owning a stake in something stranger and considerably more valuable.