Team BASE

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Team BASE

🎯 Inspiration Working with RJxTP (Racial Justice x Technology Policy) at Brandeis University, we witnessed firsthand how nonprofit teams spend 40+ hours per grant cycle on manual research, eligibility checking, and application drafting. Time that should be spent advancing racial justice was lost to administrative burden. We asked: What if AI could handle the paperwork so humans could focus on the mission?

🔨 How We Built It We developed a full-stack AI-powered grant management platform using:

React + TypeScript frontend with real-time state synchronization Lovable Cloud (PostgreSQL + Edge Functions) for persistent data and serverless APIs Google Gemini 2.5 Flash and Ollama for intelligent grant matching and partner discovery Resend API for automated stakeholder notifications The architecture follows a conversational AI paradigm—users describe what they need in natural language, and the system discovers grants, scores fitness, generates drafts, and tracks the entire pipeline from discovery to decision.

📚 What We Learned Context-aware AI beats generic search: Training the AI to understand RJxTP's specific mission produced dramatically better grant matches than keyword search Real-time sync is non-negotiable: Nonprofit teams work collaboratively; WebSocket subscriptions ensure everyone sees live updates Trust requires transparency: Users needed to see why the AI recommended each grant (fitness scores, eligibility breakdown) to trust its suggestions ⚡ Challenges AI hallucination control: Early versions invented grant opportunities. We implemented strict JSON parsing and source validation Email deliverability: Configuring Resend with proper sender verification took iteration Balancing automation vs. control: Users wanted AI assistance but not blind automation—we added human approval gates at every critical step 📈 Impact The platform reduces grant research from days to minutes, with measurable improvements:

90% reduction in discovery time 10x increase in partner leads through AI web intelligence Consistent, data-driven fitness scoring replacing gut-feel decisions