Grant consultants spend days manually searching for government grants across dozens of sources: federal databases, state portals, listing pages, open web. For each client they repeat the process, cross-referencing eligibility criteria against business profiles. The high-value grants get buried under hundreds of irrelevant ones.

I saw an opportunity to turn a manual, expensive process into a product.

Grant Finder automates the entire cycle. An operator creates a client profile once, triggers a scan across five discovery layers, and gets a ranked list of matched grants with plain-language rationale and direct links. The whole scan costs about a dollar, replacing days of consultant time. I validated the opportunity with consultants and tested with real start-ups before shipping working software.

I owned everything: problem definition, UX architecture, product spec, scoring framework, and working software. The key design decisions were all shaped by one constraint most designers never face: every API call costs money. I decoupled discovery from scoring so operators control spend.

I built pause gates between pipeline layers so users see the AI "thinking" instead of staring at a black box. I designed filter-first results with action-based triage (Apply Now, Watch List, Skip) because consultants prioritise by next action, not by score number. The two-pass AI scoring (Haiku for cheap bulk filtering, Sonnet for deep refinement of top candidates) is a product design decision disguised as architecture. Users never wait for expensive processing on grants that clearly don't match.

Interested in the story behind this project? Let's have a chat!