GovGrant AI | Devpost

Homepage
How it works
Submit Details & Get Relevant Grants
Saved Grants
Contact Page

🌟 Inspiration

Finding and understanding government grants and welfare schemes can be incredibly complex and overwhelming. Information is scattered across platforms, buried in jargon, and often hard to interpret. This creates an accessibility barrier, especially for students, farmers, small business owners, and low-income families who need this support the most.

The inspiration behind GovGrant AI was to democratize access to these vital resources using AI. By leveraging natural language understanding, we aimed to build a tool that simplifies the discovery of relevant grants and empowers people to unlock opportunities that can improve their lives and communities.

💡 What it does

GovGrant AI is an intelligent grant discovery platform designed to help users effortlessly find and understand government grants and welfare schemes using plain language.

Users can:

🔍 Search for Grants: Input their age, profession, income, country, and goals (e.g., "start a business" or "housing aid").
🤖 AI-Powered Results: Get a curated list of grants with simplified summaries generated by Gemini AI.
📄 Explore Grant Details: View eligibility criteria, summaries, and official source links.
📌 Save Grants: Bookmark interesting grants, add notes, and manage them locally.
🧭 Get Guidance: Learn about the platform’s mission and access FAQs.

The core intelligence comes from Gemini AI integrated through the Genkit framework, enabling dynamic and user-friendly natural language processing and summarization.

🛠️ How I built it

GovGrant AI was built using a modern, scalable web development stack with best-in-class tools for AI, UI, and user experience:

🧰 Technologies Used:

Frontend Framework: Next.js with App Router (React-based server-side rendering)
Programming Language: TypeScript (ensures type safety and scalability)
UI Components: Shadcn UI — highly accessible, customizable components
Styling: Tailwind CSS for rapid, utility-first UI styling
AI Integration: Genkit framework using Google Gemini models
Form Handling: React Hook Form for elegant and performant forms
Validation: Zod — for strong schema enforcement in user input and AI outputs
Icons: Lucide React for consistent and modern iconography

🧩 Development Process

✅ Set up a Next.js + TypeScript project structure
🎨 Integrated Shadcn UI and customized the visual theme using Tailwind
🧠 Built Genkit AI flows (grantSearch, grantSummarize) with Gemini
🧾 Defined input/output schemas using Zod to validate both user input and AI responses
🔁 Connected the frontend to Genkit functions for real-time responses
💾 Developed logic to bookmark and manage saved grants using local storage
🔬 Iteratively tested and refined UI, AI prompts, and overall experience

🚧 Challenges I ran into

Prompt Engineering: Designing prompts that yielded structured, schema-compliant results from Gemini was tricky. Handling null results gracefully also required prompt and UI tweaks.
Data Consistency: The AI sometimes returned improperly formatted or overly verbose data. We had to enforce strict schemas using Zod and sanitize results.
Async Logic and UI: Managing loading states, syncing results, and handling no-data scenarios in a seamless way was a key frontend challenge.
Component State: Managing saved state (notes, bookmarks) across components required clean architecture and state management.
Limited Free APIs: Without access to government APIs in some regions, we had to rely solely on AI for intelligent scraping and summarization.

🏆 Accomplishments I'm proud of

🧠 Successfully built a functional AI-powered discovery tool using Genkit + Gemini
🌐 Designed and implemented a clean, responsive, and accessible UI
🔍 Created a dynamic, personalized search experience
💾 Implemented bookmarking and grant management in-browser
📦 Defined robust schemas and prompt flows to ensure consistent AI responses

📚 What I learned

🔄 Real-world application of Genkit and Gemini models for text summarization and search
🧩 Importance of clear schema validation when working with AI data
🧠 Skills in prompt design and debugging natural language model behavior
⚛️ Deepened understanding of Next.js App Router, Server Components, and React state handling
🔧 Reinforced principles of iterative development, from prototyping to production

🚀 What's next for GovGrant_AI

🔗 Integrate official APIs (India’s MyScheme, US Grants.gov, EU funds, etc.) for verified and real-time grant listings
🧹 Advanced Filters & Sorting (by deadline, amount, sector, keywords)
👤 User Profiles & Personalization for saving searches, preferences, and syncing across devices
🔔 Notifications for new grants or approaching deadlines
📝 Application Helpers (e.g., AI-generated templates, proposal writing support)
🌐 Multi-language support to scale globally
📊 Feedback Loop for users to rate results and improve relevance over time
📄 Grant PDF parsing and document understanding using AI