OPEN FOR REQUESTS  //  GEARY_WORKS · PORTFOLIO
X:500 Y:500
14 LIVE · 3 BUILDING 00:00:00
All projects
Live since May 29, 2026 Project 12 / 17

Reckon.

A free 70-question founder archetype assessment that tells you what to avoid.

Status
Live
Launched
May 29, 2026
Pricing
Free
Stack
Next.js · TypeScript · Postgres · Claude · Resend
Category
Web apps & tools AI-powered
Reckon home page at reckon.report
reckon.report

What it is

Reckon is a free founder archetype assessment at reckon.report. Seventy questions score you on eight psychometric dimensions — the Big Five plus risk tolerance, locus of control, and need for achievement — sort you into one of eight archetypes, and generate a long written report on your operating style, your blind spots, and the kinds of businesses you should probably refuse to start.

Who it's for

Anyone about to start a business, or already running one, who wants an honest read on their own operating style before betting money and years on it. It's built for founders and want-to-be founders, not HR departments running team assessments.

Why I built it

Most personality tests either cost nothing and tell you nothing useful — generic Big Five percentiles with no application — or cost a fortune and sit locked behind certified consultants, like Hogan or Predictive Index. Nothing sits in between: a self-serve assessment built on real psychometric instruments that actually speaks to the decisions founders face. Reckon is meant to be that missing option.

How it works

You answer 70 questions — a mix of forced-choice scenarios, values trade-offs, rankings, and rating items — in about 30 minutes. The scoring is built on established psychometric instruments (NEO-PI-R, BFI-2, DOSPERT, Rotter I-E, McClelland), not an invented framework. Claude writes a 20-plus section report from your scores, including the gap between who you say you are and who your answers say you are. It's free start to finish — there's no payment anywhere in the product. Email capture is the point: it's a lead magnet for a future product, not a paid report.

What's next

The scoring currently runs on placeholder calibration until enough real completions accumulate; once that threshold is hit, it switches from descriptive bands (high, moderate, low) to real population percentiles. Distribution is still an open question — the next step is a small, deliberate launch to get real signal.

Related

IF THIS, THEN THESE.

Same category, same maker.

WANT SOMETHING
LIKE THIS?

I build and ship software products end to end, fast, with AI. Want one built, or want to learn to do this yourself? Email me.