How an AI Agent and Automation Took Grant Feedback from 15 Weeks to 2 Days

At a glance

• Client: State economic development agency
• Challenge: Grant applicants waited up to 15 weeks for early feedback
• Solution: Login-free, automated intake on Microsoft Power Platform, plus a governed, conversational AI agent on AWS
• Result: Feedback in 2 days, and real-time guidance to the right program

Overview

Communities looking for funding to grow their local economies now get a clear answer in days instead of months. Their state's economic development agency also showed how government can put generative AI in front of the public responsibly, building a foundation other programs and public services can grow into.

The agency awards more than 300 grants a year to municipalities, nonprofits and community organizations. Before applying, communities can submit an Expression of Interest for early guidance on which programs fit their project. But that step meant logging into a legacy system, working through a complex form, and waiting up to 15 weeks for staff to review it by hand.

The agency wanted applicants to get answers faster and with less effort, and it wanted the new experience ready for a major statewide gathering of municipal leaders.

How We Helped

1. Timed for the statewide event. Using a crawl-walk-run approach, a public-facing platform on Microsoft Power Platform was time-boxed to be ready for the annual gathering of municipal leaders.
2. No login, no barrier. Applicants submit without logging in, with no limit on submissions, and receive a feedback report by email.
3. Faster reviews for staff. Feedback reports are generated automatically, so staff can review each submission and its feedback quickly instead of writing reports by hand.
4. Better every grant cycle. Since 2023, the platform and feedback report have been refined with each cycle.
5. An AI guide for applicants. Applicants describe a project in their own words. The AI agent shows available programs by town and organization, summarizes eligibility, suggests referrals, and can submit the Expression of Interest for them.
6. Governed from the start. A persistent audit log captures each conversation and the basis for the agent's recommendation, and human-validation points were defined before release for higher-consequence outputs. The chat experience follows the state's design system and its accessibility and security standards.

Results

Before and after

• Feedback: up to 15 weeks → 2 days
• Access: log in to a legacy system → no login, no submission limit
• Feedback reports: written by hand → generated automatically and emailed
• Program guidance: wait for staff → real-time recommendations from an AI agent

The result is a faster, more transparent and more accessible experience for applicants, and a more efficient way of working for agency staff. With an audit trail and human validation built in, the AI agent shows how a state agency can put generative AI in front of the public responsibly.

Capabilities

• Grants management modernization
• Low-code and automated workflows
• Applied and generative AI for the public sector
• Responsible AI governance
• Accessible, citizen-facing digital services
• Embedded, multi-year managed delivery

Built with: Microsoft Power Platform (Power Pages, Power Apps, Power Automate, Dataverse) · AWS Bedrock AgentCore and Strands Agents with Anthropic Claude Sonnet · AWS Lambda, API Gateway, DynamoDB, S3 · React and TypeScript

Have a public-facing process that keeps people waiting? Connect with our team about where automation and AI can help.

‍