GrantFlow.AI

Intelligent grant management platform that automates discovery, planning, and application workflows for researchers and institutions.

What is GrantFlow.AI?

GrantFlow.AI is a comprehensive platform designed to streamline the grant management process. Researchers and institutions can discover relevant grant opportunities, organize collaborative planning, and generate high-quality grant applications—all in one place.

Our platform combines intelligent document processing, real-time collaboration, and AI-powered content generation to transform how teams approach grant management. Whether you're tracking opportunities, coordinating across departments, or drafting applications, GrantFlow.AI provides the tools to work smarter.

Key Features

• Intelligent Grant Discovery: Automated monitoring of grant opportunities with targeted notifications
• Collaborative Planning: Real-time document editing with team members using CRDT-based synchronization
• AI-Powered Generation: Leverage RAG technology to generate grant applications and materials
• Document Processing: Advanced PDF, DOC, and web content extraction with semantic indexing
• Multi-Tenant Organization Management: Role-based access and team collaboration
• Vector-Powered Search: Semantic search across your documents and grants
• Integration-Ready: Built-in support for Discord notifications and custom webhooks

Architecture

System Architecture

GrantFlow.AI is built as a full-stack monorepo with a modern frontend, microservices backend, and real-time collaboration layer:

graph TB
    subgraph Frontend["Frontend Layer"]
        A["Next.js 15 Web App<br/>(React 19, Tailwind)"]
    end

    subgraph Collab["Real-Time Collaboration"]
        E["Hocuspocus<br/>WebSocket Server"]
        F["Y.js CRDT<br/>Document Sync"]
    end

    subgraph API["API Gateway & Services"]
        B["Litestar<br/>REST API"]
        C["Pub/Sub<br/>Message Bus"]
    end

    subgraph Services["Microservices"]
        D1["Document<br/>Indexer"]
        D2["Web<br/>Crawler"]
        D3["RAG<br/>Engine"]
        D4["Grant<br/>Scraper"]
    end

    subgraph Data["Data Layer"]
        G["PostgreSQL<br/>+ pgvector"]
        H["GCS<br/>Storage"]
        I["Vector<br/>Database"]
    end

    A -->|HTTP/WebSocket| B
    A -->|Real-time Sync| E
    E -->|Y.js Protocol| F
    B -->|Publish Events| C
    C -->|Subscribe| D1
    C -->|Subscribe| D2
    C -->|Subscribe| D3
    D1 --> G
    D1 --> H
    D1 --> I
    D2 --> G
    D3 --> I
    D4 --> G
    B --> G
    B --> I

Loading

RAG Pipeline

The Retrieval Augmented Generation pipeline powers intelligent content generation using document context and domain expertise via Kreuzberg:

graph LR
    subgraph Input["Input Stage"]
        A["Blueprint<br/>Requirements"]
    end

    subgraph Processing["Processing Stage"]
        B["Query<br/>Planning"]
        C["Vector<br/>Retrieval"]
        D["Kreuzberg<br/>Context"]
    end

    subgraph Synthesis["Synthesis Stage"]
        E["Multi-Model<br/>LLM"]
        F["Content<br/>Generation"]
    end

    subgraph Output["Output"]
        G["Grant<br/>Application<br/>Materials"]
    end

    A -->|User Intent| B
    B -->|Semantic Query| C
    C -->|Relevant Docs| D
    D -->|Domain Knowledge| E
    E -->|LLM Processing| F
    F -->|Output| G

Loading

Real-Time Collaboration

Concurrent document editing is powered by CRDT (Conflict-free Replicated Data Type) synchronization with Y.js and Hocuspocus:

sequenceDiagram
    participant User1 as User 1<br/>Client
    participant User2 as User 2<br/>Client
    participant Server as Hocuspocus<br/>WebSocket Server
    participant YJS as Y.js Document<br/>State

    User1->>Server: Connect & subscribe to doc
    User2->>Server: Connect & subscribe to doc
    Server-->>YJS: Load persisted state

    User1->>YJS: Edit text (local CRDT op)
    YJS-->>Server: Broadcast update
    Server-->>User2: Send update (WebSocket)
    User2->>YJS: Apply remote change

    User2->>YJS: Edit different section
    YJS-->>Server: Broadcast update
    Server-->>User1: Send update (WebSocket)
    User1->>YJS: Apply remote change (auto-merge)

    Note over User1,User2: No conflicts - CRDT handles concurrent edits

Loading

Technology Stack

Layer	Technology
Frontend	Next.js 15, React 19, TypeScript, Tailwind CSS, Zustand
Real-time Collab	Hocuspocus, Y.js CRDT, WebSocket
Editor	TipTap (rich text with collaborative capabilities)
API	Litestar (Python async framework)
Authentication	Firebase Auth
Database	PostgreSQL 17 with pgvector extension
Processing	Python microservices (Indexer, Crawler, RAG, Scraper)
AI/LLM	OpenAI, Anthropic, Vertex AI
Storage	Google Cloud Storage
Messaging	Google Cloud Pub/Sub
Infrastructure	Google Cloud Run, OpenTofu/Terraform
Observability	OpenTelemetry, Cloud Logging

Quick Start

This project is a monorepo with multiple services. For detailed setup instructions, build procedures, and development workflows:

See CONTRIBUTING.md for:

• Prerequisites and environment setup
• Local development commands
• Testing procedures
• Deployment guidelines

Documentation

• Technical Docs: Architecture, API specs, and security documentation
• Contributing Guide: Development setup and workflows
• Backend Services: Individual service READMEs
• Packages: Database, utilities, and shared code
• Frontend: Next.js application guide
• Editor: Collaborative editor package

Repository Map

Core Infrastructure

• /.github - GitHub Actions CI/CD workflows and reusable actions
• /terraform - Infrastructure as Code using OpenTofu
  • environments/ - Environment-specific configurations (staging, production)
  • modules/ - Reusable Terraform modules
• /functions - Serverless monitoring & automation
  • app_hosting_alerts/ - Firebase deployment notifications
  • auth_blocking/ - User authentication validation
  • budget_alerts/ - Cost monitoring alerts
  • email_notifications/ - Transactional email service

Frontend Applications

• /frontend - Next.js 15 web application
  • Modern React 19 with TypeScript
  • Tailwind CSS for styling
  • Firebase Authentication
  • Zustand state management
• /editor - TipTap collaborative editor package
  • Rich text editing capabilities
  • Real-time collaboration support
• /crdt - CRDT server for real-time collaboration
  • Hocuspocus WebSocket server
  • Y.js document synchronization

Backend Services

• /services/backend - Main API service
  • Litestar async framework
  • JWT authentication with Firebase
  • PostgreSQL with SQLAlchemy 2.0
  • Organization-based multi-tenancy
• /services/indexer - Document processing pipeline
  • PDF/DOC/HTML extraction
  • Chunk generation and embeddings
  • Vector database indexing
• /services/crawler - Web content extraction
  • Intelligent link following
  • Content extraction and cleaning
  • Rate limiting and robots.txt compliance
• /services/rag - AI-powered content generation
  • Retrieval Augmented Generation
  • Multi-model support (OpenAI, Anthropic, Vertex AI)
  • Grant template and application generation
• /services/scraper - Grant opportunity discovery
  • NIH grants.gov integration
  • Automated opportunity monitoring
  • Discord notifications

Shared Packages

• /packages/db - Database layer
  • SQLAlchemy models and migrations
  • Alembic migration management
  • Database utilities and helpers
• /packages/shared_utils - Common utilities
  • AI/LLM integrations
  • GCS storage operations
  • Pub/Sub messaging
  • OpenTelemetry instrumentation
  • Embeddings and NLP utilities

Documentation and Testing

• /docs - Comprehensive technical documentation
  • Architecture diagrams
  • API specifications
  • Security documentation
  • Deployment guides
• /e2e - End-to-end testing utilities
  • Monitoring test helpers
  • Webhook testing tools
  • Integration test fixtures

Development Tools

• /.vscode - VS Code workspace settings and recommended extensions
• /.idea - IntelliJ IDEA project configuration
• /scripts - Utility scripts for development and deployment
• Taskfile.yaml - Task automation (replacement for Makefiles)
• docker-compose.yaml - Local development environment

Contributing

We welcome contributions! Please see CONTRIBUTING.md for:

• Code standards and conventions
• Testing requirements
• Pull request process
• Development setup

License

GrantFlow.AI is dual-licensed under the Business Source License 1.1 (BSL 1.1) and the Apache License 2.0.

• Under BSL 1.1: You may use this software for any purpose except offering it as a managed service to third parties without a commercial license
• Conversion Date: January 10, 2030 - The software will automatically convert to Apache 2.0 four years after the release date, making it fully open source

For detailed license terms, see LICENSE.md.