Inspiration
We saw how much time nonprofits and early-stage teams waste navigating the grant process. Organizations like New Roots Urban Farm are doing important work but spend days searching for grants, checking eligibility, and rewriting similar applications. We wanted to simplify that process and make funding more accessible, especially for smaller, hyper-local teams without dedicated resources.
What it does
GrantMatch.ai is a hyper-local grant matcher that turns a single sentence into a complete funding workflow. It finds relevant grants, checks eligibility with clear blockers and readiness scores, generates a tailored document checklist, and drafts a ready-to-submit application, all in under 60 seconds.
How we built it
We built GrantMatch.ai as a multi-agent system using Gemini and K2-Think, orchestrated through a FastAPI backend with live streaming via SSE. A profile parser structures user input, a grant finder uses Google Search with a fallback dataset, a reasoning model evaluates eligibility across grants, and a grant writer generates full application sections in parallel. We added constrained JSON outputs, retry logic, and parallel execution to make the system fast and reliable.
Challenges we ran into
Ensuring reliability across multiple LLM calls was a major challenge, especially handling inconsistent outputs and rate limits. We also had to carefully manage dependencies between agents while still keeping the system fast through parallel execution. Another challenge was making eligibility checks accurate and grounded in real requirements rather than generic reasoning.
Accomplishments that we're proud of
We built a working end-to-end system that goes from raw input to structured profile, ranked grants, eligibility analysis, document checklist, and a full draft in real time. The system streams each step live and demonstrates a clear use case, showing how Grain can reduce a multi-day process into minutes.
What we learned
We learned that the hardest part of grants is not discovery but understanding eligibility and preparing materials. Structuring data early makes everything downstream more reliable. We also saw that users trust the system more when they can see how decisions are made, especially around eligibility and blockers.
What's next for GrantMatch.ai
Next, we want to expand our grant database, improve eligibility checks using more structured requirements, and support full submission workflows. We also plan to add document autofill, deadline tracking, and collaboration features so teams can manage the entire funding process in one place.
Log in or sign up for Devpost to join the conversation.