Skip to content

Ai4impact co-op | Fall 2023- Spring 2024 project

GrantWell

Simplifying Federal Grant Applications with AI

ABOUT THE PROJECT

https://drive.google.com/file/d/1zlkh5CmRd8uXFSewLboM_eagaLmFqeoY/view?usp=sharing

Assisting Massachusetts municipalities, environmental justice groups, rural towns, and historically-underserved communities to apply for federal and state grant, grant-matching, and tax-break opportunities. The tool allows municipalities to upload a grant application and receive a clear, concise summary of grant requirements in order to determine eligibility. The tool also features an AI-powered chat interface that assists users in crafting compelling narratives tailored to each grant. This supports both higher volume and higher quality applications.

Partner: Massachusetts Executive Office of Administration and Finance

Status: Fully functional and piloted with municipalities across the state as part of the Director of Federal Funds roadshow.

Impact: If fully deployed, it could enable Massachusetts municipalities to secure millions more in federal funding each year, supporting clean energy projects, road repairs, and affordable housing developments.

PARTNERS

Executive Summary

  • Challenge: Underserved municipalities and community groups often lack the staff, technical capacity, or time to navigate complex federal grant applications, perpetuating funding inequities.
  • Solution: GrantWell is a public, AI-enabled platform that streamlines the grant application process with NOFO summarization, AI-powered writing assistance, and guided grant workflows.
  • Impact: Grant writers can more quickly assess eligibility, craft stronger narratives, and stay on track with deadlines. Municipalities now have access to a grant support tool tailored to their capacity and needs.
  • Partners: Massachusetts Federal Funds and Infrastructure Office, Burnes Center for Social Change

Problem Context

  • Background: Many local governments struggle to keep up with federal funding opportunities due to staffing shortages and bureaucratic complexity. Across Massachusetts, communities face urgent infrastructure and public health challenges—from over 5800 potholes reported in Boston to dangerous lead pipes still in service in Medford. These persistent issues highlight the clear need for timely, well-funded interventions.
  • Urgency/Need: The federal government allocates billions annually for which Massachusetts is eligible for $ 17.5 billion in competitive grant funding, yet less than 30% of that is successfully accessed by many local governments. Increasing local capacity to apply for and win grants is essential to unlocking this funding and addressing critical health, safety, and infrastructure needs.
  • Target Audience: Municipal grant writers, technical assistance providers, and nonprofit organizations across Massachusetts.

Innovation Process

  • Approach: Agile development with iterative user feedback, grounded with human-centered design.
  • Co-Creation: Users were engaged via live demos during regional roadshows. Feedback from over a dozen  municipalities, and stakeholders  directly informed feature prioritization and UX improvements.
  • Data Sources: Public NOFOs from Grants.gov; user-uploaded documents (e.g., PDFs, CSVs); structured summaries stored in Amazon S3.

AI Solution Overview

GrantWell is a cloud-native web platform that helps municipalities discover relevant funding opportunities, understand complex NOFO requirements, and generate high-quality grant narratives using generative AI. It is built on a modern serverless architecture using Amazon Web Services (AWS) and a React-based frontend.

Key Features:

  • Grant Recommendation Assistant: This feature uses natural language inputs from users to suggest relevant NOFOs, leveraging vector search over indexed metadata.
  • NOFO Summarization: When a NOFO is uploaded (manually or via the Grants.gov scraper), AWS Lambda functions process the PDF and invoke Claude 3.5 (via Amazon Bedrock) to extract and categorize key information—Eligibility, Required Documents, Narrative Components, and Key Deadlines. The structured summaries are stored in Amazon S3 and served via API Gateway.
  • AI Chat Assistant: A React-based chat interface allows users to query NOFOs, ask eligibility questions, and receive section-specific writing support. Chat requests are routed through AWS API Gateway to backend Lambda functions, which retrieve context from S3 and invoke Claude via Bedrock. Uploaded user files (PDFs, DOCs, CSVs) are grounded into responses using retrieval-augmented generation (RAG) techniques.
  • Guided Grant Writer: A multi-step workflow guides users through project basics and a custom questionnaire. Collected inputs are processed into structured prompts that drive the AI drafting logic, returning narrative content section by section.

Admin Dashboard + NOFO Scraper: Admins can upload NOFOs directly or rely on a scheduled pipeline that fetches new NOFOs daily from Grants.gov. A JSON-based index prevents duplication. Uploaded or fetched NOFOs are stored in Amazon S3 and indexed for search and summarization.

Outcomes & Impact

  • Quantitative Results:
    • Estimated time saved per application: 50–70%
    • Doubled the grant submission rate
    • Processed and summarized 10+ NOFOs for pilot communities
  • Qualitative Feedback:
    • “This is exactly what I need. Understanding the NOFOs is what takes the most time.”
    • Users appreciated the clarity and accessibility of the summaries and drafting guidance.

Lessons Learned

  • What Worked: Live demos were key to engagement. NOFO summarization and guided drafting were widely praised for reducing user burden.
  • What Didn’t: Some users misunderstood the chatbot as a generic assistant, highlighting the need to clarify its domain-specific design.

Adaptability: The tool can be extended to other states or domains (e.g., housing, climate, education) with minimal configuration.

Future Roadmap

  • Include post award compliance
  • Calendering tools to track deadlines

Acknowledgments

  • Project team: Anjith Prakah Chathan Kandy, Jai Surya Kode, AI For Impact
  • Stakeholder collaborators: Quientin Palfrey, Robert (Bob) LaRocca, Mehar Jauhar.
  • Organizational partners: Massachusetts Federal Funds and Infrastructure Office, The Burnes Center for Social Change at Northeastern University

Link to Source Code

https://github.com/The-Burnes-Center/AI4Impact-GrantWell

PROJECT TEAM

  • Shreya Thalvayapati
  • Serena Green
  • Deepikasai Mettu

We use cookies to improve your experience on our sites. By continuing to use our sites, you agree to our Privacy Statement.