Granted

Inspiration

We wanted to build a system where funding finds the NPO, shifting the paradigm from "Pull" (active checking) to "Push" (proactive notification)

Granted is an automated grant discovery engine that replaces manual hunting with hyper-relevant curation.

Interactive Profiling: instead of filling boring forms, NPOs chat with our bot. It "interviews" them to understand their mission, budget, and KPIs, converting vague descriptions into a structured data profile.
Curator: based on a scheduling agent, the system sends a high-value digest. It doesn't just simply list grants; it ranks them using vector similarity and quantum matching (checking if the grant amount fits the project size)
Detection: our system never sleeps. As soon as a new grant drops on the OurSG portal, our system detects it and instantly "pings" only the specific NPOs that are a match with a "Flash" notification

Frontend/Dashboard: Next.js for the NPO interface.

Database: A schema tracking Profile_JSON, Vector_Embeddings, and Frequency preferences.

Notifications: Resend and React Email to generate clean, mobile-responsive summaries ("Top 3 Matches").

Data Unstructuredness: Transforming vague NPO "About Us" pages into the structured JSON required for the users' database schema was difficult. Tuning the LLM to accurately extract "Average Project Budget" from conversational text took iteration.
Balancing "Relevance" vs. "Noise": Initially, the vector matching might be too broad. We had to refine the Scoring Logic to ensure we didn't spam users, focusing strictly on high-confidence matches (The "Top 3").
Real-time Reliability: Implementing the "Delta Detection" to reliably catch changes on the OurSG portal without false positives.

True Automation (with control): Successfully built the "Push" architecture where a new grant triggers a tailored email to the right NPO either automatically without human intervention or when the user triggers send.
Zero Missed Deadlines: Implementing the deadline proximity boost to ensure NPOs are alerted before it's too late.
Conversational UX: Replacing the industry-standard generic forms with a conversational "interview" that feels personal and intelligent.

Automated Drafting: Moving beyond just finding the grant to helping write the first draft of the proposal using the stored NPO profile.
Multi-Source Aggregation: Expanding the "Watchdog" to monitor corporate foundations and private philanthropy, not just government portals like OurSG.
Feedback Loop: Implementing a system where NPOs can "thumbs up/down" matches to further fine-tune their Vector Embedding for future recommendations.