Inspiration

The TSAO Foundation in Singapore works tirelessly to support aging and community care initiatives, but like many social impact organizations, they face a significant challenge: discovering relevant grant opportunities in a fragmented landscape. Grant information is scattered across government portals, foundation websites, and various databases, making it time-consuming and difficult for organizations to identify funding that aligns with their mission.

We were inspired to create GrantMatch AI after learning about the countless hours nonprofit teams spend manually searching for grants, often missing opportunities simply because they didn't know they existed. We wanted to leverage artificial intelligence to democratize access to funding information and help organizations like TSAO Foundation maximize their impact.

What it does

GrantMatch AI is an intelligent Agentic AI grant discovery platform that automatically matches organizations with relevant funding opportunities.

The system analyzes an organization's profile, mission, programs, and past projects, then uses natural language processing and machine learning to identify grants that align with their work.

Key features include:

  • Smart Matching Algorithm — Uses AI to understand both grant requirements and organizational capabilities
  • Real-time Opportunity Tracking — Continuously monitors sources for new grant announcements
  • Eligibility Scoring — Provides a compatibility score for each grant opportunity
  • Caregiver CRM — Integrated relationship management for eldercare organizations

How we built it

We developed GrantMatch AI using a modern tech stack designed for scalability and intelligent processing:

Layer Technology
AI/ML Google Gemini 1.5, LangChain, Semantic Vector Search
Backend Python, FastAPI
Frontend React
Data Collection Playwright (browser automation), Web scraping
Database ChromaDB (vector store), JSON
Deployment Docker

Core AI Capabilities:

  • Natural Language Processing — Implemented transformer models to understand grant descriptions and organizational profiles
  • Multi-Agent System — Specialized AI agents for search, matching, eligibility checking, and proposal writing
  • RAG (Retrieval-Augmented Generation) — Grounds AI responses in real grant data
  • Web Scraping & APIs — Automated data collection pipelines to aggregate grant information from multiple sources

Challenges we ran into

🕷️ Web Scraping Government Portals

Singapore's OurSG Grants website uses JavaScript rendering and rate limiting, requiring Playwright browser automation instead of simple HTTP requests. We encountered 429 rate limit errors and had to implement delays and retry logic.

📋 Incomplete Grant Data

Scraped grants often had missing fields (applicant type, categories, funding amounts). We solved this by developing a keyword-based auto-categorisation system that analyses grant descriptions to infer missing metadata.

⏱️ Slow AI Response Times

The Gemini API takes ~30-60 seconds to score 60+ grants against an organisation profile. We implemented 1-hour caching to ensure fast subsequent page loads while keeping scores fresh.

🐳 Production Deployment

Getting the React frontend and FastAPI backend to work together in a single Docker container required careful static file serving setup and proper path handling for SPA routing.

🔍 Filter Functionality

Case sensitivity issues between frontend dropdown values and backend data caused filters to fail silently. Debugging required tracing data through the full stack.


Accomplishments that we're proud of

✅ Built an end-to-end agentic AI system that goes beyond keyword search to reason about grant fit, eligibility, and relevance

Centralized grant intelligence — Transformed scattered, hard-to-navigate government grant data from Singapore's OurSG portal into a clean, searchable, AI-powered database

✅ Designed an eligibility scoring mechanism that helps nonprofits quickly prioritize high-value, high-fit grants instead of manually reviewing dozens of listings

✅ Applied AI to a real social problem, grounding the solution in the lived challenges of organizations like TSAO Foundation rather than a purely theoretical use case

✅ Delivered a scalable architecture, making the platform extensible to other nonprofits, sectors, and regions

Caregiver CRM integration — Beyond grants, we built a caregiver management system with profile tracking and care status management for the Tsao Foundation's eldercare mission


What we learned

"Discovery is the real bottleneck, not application writing. Nonprofits often miss funding simply because they don't know it exists."

Key insights:

  1. Context matters more than keywords — Understanding mission statements, program goals, and grant intent requires semantic and contextual analysis, not basic matching

  2. Agentic AI adds real value when it actively monitors, evaluates, and prioritizes opportunities instead of acting as a passive search tool

  3. Trust and transparency are critical in AI systems for social impact — Users need to understand why a grant is recommended, not just see a score

  4. Building for nonprofits requires empathy — Time, manpower, and technical resources are limited, so simplicity and clarity matter


What's next for GrantMatch AI

Feature Description
📧 Automated Monitoring Email/SMS alerts when new grants matching the organization's profile are published or when deadlines approach
📊 Application Tracking Track grant application status from draft → submitted → awarded, with document management and deadline reminders
👥 Multi-Organization Support Allow multiple non-profits to use the platform with their own profiles and grant matches
🎯 Success Prediction Analyze historical grant awards to predict likelihood of success for specific applications
🌏 Expanded Data Sources Integrate grants from corporate foundations, international organizations, and regional government programs beyond Singapore
✍️ Collaborative Proposals Team-based proposal editing with AI suggestions, version history, and approval workflows
🚀 Global Expansion Adapt the platform for regional and global grant ecosystems

Tech Stack

┌─────────────────────────────────────────────────────────────┐
│                     GrantMatch AI                           │
│              AI-Powered Grant Discovery                     │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│   ┌─────────┐    ┌─────────┐    ┌─────────┐               │
│   │ Search  │ →  │ Analyze │ →  │ Generate│               │
│   │   🔍    │    │   🤖    │    │   📄    │               │
│   └─────────┘    └─────────┘    └─────────┘               │
│                                                             │
│   AI Layer:     Google Gemini • Vector Search • RAG        │
│   Backend:      Python • FastAPI                           │
│   Frontend:     React                                      │
│   Data:         ChromaDB • Playwright Scraping             │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Built with ❤️ for Hack4Good 2026 — Empowering TSAO Foundation's mission to support eldercare in Singapore

Built With

Share this project:

Updates