Two hackathons, two honest results

I will report both hackathons the same way: with the numbers.

Operario went to Frontier and finished third out of 25 teams, with 45 out of 56 points — 80 percent, ahead of 88 percent of the field. The challenge median was 23 percent; the best team scored 93 percent. On the rubric, Systems Design came in at 2.7 out of 4 and Execution at 3 out of 4. Every delivery was on time. Operario is an appointment-taking system for local businesses with real admin burden — vet clinics and auto shops first, places running old Windows systems with no API. The architecture is deliberately conservative: an untrusted proposer that takes one action per screenshot, a judge at the gates, and drafts rather than sends. Privacy is the first gate, not an afterthought.

Nury went to Gloo and made the Top 25 finalists. It did not win the finals. I am writing that plainly because an earlier note of mine got it wrong, and the correction belongs in the record too. Nury is a 3Metas project, and it stays in 3Metas. The finals asked for a 90-second video by 9 a.m., and that constraint was the real lesson: if you cannot explain the product in 90 seconds, you do not understand it yet.

What the two results have in common is the part I care about. Execution beat novelty in both rooms. Shipping on time, with a system you can explain and defend, outperforms a cleverer idea that is half built. That is also how I build products now: small, verifiable steps, a judge at the gate, and results reported exactly — including the ones that are not wins.