Match the grant thesis
Every grant program has a thesis, even when the page looks broad. Some want open source tools, some want ecosystem adoption, some want safety work and some want startups using a platform.
Do not send the same proposal everywhere. Rewrite the first paragraph so the funder can see why your project belongs in their program.
Write a milestone, not a dream
A grant application should explain what will exist at the end of the funding period. Name the demo, repo, dataset, benchmark, deployment or user group.
Avoid vague promises like build an AI platform. Use concrete outputs such as release a retrieval evaluation harness for legal documents with ten public test cases.
Make the budget believable
A simple budget is better than a theatrical one. List compute, data, contractor help, design, user research or living expenses if the program allows it.
The amount should match the work. Asking for too little can look unserious, but asking for too much without a milestone makes the project look unplanned.
Show evidence before asking
A prototype, GitHub repo, waitlist, benchmark, prior hackathon demo or user interview can de-risk the application quickly.
If you have no evidence yet, build the smallest proof first. Grants are easier to win when the reviewer is funding momentum, not imagination.
< read by a human · updated as things change >
browse hackathons