Prezenti AI Builder Sponsorships (Trial)

How you are assessed

Your public code activity is scored against a fixed rubric. Four signals carry most of the weight.

Origination. You create things that did not exist. Forking is free; originating is not.
Finishing. Most side projects die around commit three. Releases, documentation, a deployed URL — the last ten percent, the part with no dopamine in it.
Cadence. You keep showing up. Distinct weeks of activity, which is much harder to inflate than a commit count.
Acceptance. Other people merged your work, across several different projects.

Stars and follower counts are not scored at any weight. They measure access, and access is exactly what this is built to look past.

You can check our work

The rubric is published in full, along with the code that applies it. Clone it and reproduce your own score:

git clone https://github.com/prezenti/talent-engine
cd talent-engine
python3 -m talent_engine.cli score \
    --program prezenti-sponsorship-trial --handles YOUR_GITHUB_HANDLE

If a number looks wrong to you, you can show us exactly where — which is the point of publishing it. Every score comes with linked evidence for each component; a score with no evidence behind it cannot be produced.

A score is not a decision

The automated score produces a shortlist. People decide, after reading the evidence and talking to you. No one is funded or refused by a number.

The checks that flag manipulated profiles are published alongside the rubric — including an honest account of what they do not yet catch. We would rather tell you that than imply a precision the method does not have.

What we collect

Public code activity only. Nothing about your background is inferred — there is nowhere in the system to record an inferred trait. Anything about your circumstances enters because you chose to tell us. Your contact details are stored separately from everything used for assessment, and never appear in an assessment record.

The terms

5 places, 4 months, $1,400 each ($7,000 total): Claude Max 20x, ChatGPT Pro 5x, flexible allowance
give-back 2% of revenue actually received by the product through Celo, and any grant, prize and retro-funding income won with the sponsored work. Capped at $14,000, expiring 36 months after the term, pro-rated by months received
owed to Prezenti, and to nobody else
Prezenti separately routes half of what it receives onward to Celo Community Fund — equivalent to 1% of covered income, and a commitment made by Prezenti, not by you
monthly receipts for reimbursement; monthly public updates; 30 days inactive triggers review with a 7-day cure period
continued months three and four require a month-two Celo deployment, integration or material contribution
equity taken: none; right of first offer on a future round
enforcement: reputational
payment route: direct to the verified Prezenti Safe; no 0xSplits collector
acceptance requires a public Celo EAS pledge naming your GitHub handle

Canonical terms: prezenti-sponsorship-trial-2026-08-18-v4 terms-version: 601b9cc0a959.

These are enforced by the code that runs this programme: a policy that takes equity, leaves the give-back uncapped, lets it run forever, or fails to state what we owe you will not load, and the service will not start.

Apply