Streamline investor discovery, pitch generation, and outreach with AI-powered precision.
TL;DR: Instantly match with ideal investors, craft compelling pitches and emails, and track your outreach—all in one place.
- 🥇 Most Impactful Award – AgentHacks 2025
Selected from 180+ submissions at AgentHacks 2025, hosted by Dex, AfterQuery, and AGI House. This award recognized PitchSense for its real-world relevance, agentic architecture, and polished execution under 24 hours.
- Executive Summary
- The Challenge
- How It Works
- Innovation Highlights
- Live Demo & Visuals
- Continuous Development Roadmap
- Installation & Quick Start
- Built With
- Project Structure
- Contributors
- License & Acknowledgments
PitchSense addresses the inefficiencies of traditional fundraising. By leveraging AI, it accelerates investor matching, generates tailored outreach materials, and manages communication pipelines end-to-end. The result: faster outreach, higher relevance, and better outcomes for founders.
PitchSense was originally developed at AgentHacks 2025, a leading hackathon focused on responsible, agentic AI systems. The event was hosted by Dex and AfterQuery—both Y Combinator-backed startups—alongside AGI House, bringing together over 500 attendees to tackle real-world problems with intelligent agents. PitchSense was awarded Most Impactful Project for its practical relevance, clean UX, and strong technical execution.
This project was developed as part of the research track Interfaces for Human-AI Collaboration — focused on designing HCI interactions where AI doesn’t just automate, it collaborates.
Traditional fundraising involves manual investor discovery, generic pitches, and fragmented outreach tracking. This inefficiency wastes valuable founder time and resources, reducing their focus on core business growth.
-
Investor Matching Engine
- Captures founder details (sector, stage, traction)
- Uses semantic search with OpenAI Embeddings and FAISS
- Provides ranked investor recommendations
-
Dynamic Pitch Generator
- Customizes pitch decks and emails using GPT-4
- Provides confidence tags for human oversight
-
Outreach Management
- CRM-lite tracking system (Airtable/Notion integration)
- Status indicators: Contacted, Replied, Intro Requested
- Smart Matchmaking: Highly relevant, personalized investor recommendations
- AI-enhanced Communication: Human-like tone, avoiding robotic phrasing
- Integrated Outreach Tracking: Real-time CRM synchronization for transparency
- Human-AI Collaborative Interface (HCI): Seamlessly integrates human judgment with AI suggestions, enhancing decision-making and trust.
PitchSense is more than a tool—it's your AI fundraising agent, continuously evolving through advanced AI and human-AI collaboration:
-
Phase 1: Intelligent Automation (Current)
- Robust investor matching and pitch generation
- Initial CRM integration for seamless outreach management
-
Phase 2: Enhanced Human-AI Collaboration (Next)
- Interactive feedback loops to refine AI predictions
- Real-time human-in-the-loop interventions for critical decisions
- Enhanced UI/UX design for intuitive human-AI interaction
-
Phase 3: Fully Autonomous Agentic System (Future)
- Autonomous management of investor communications
- Predictive analytics to forecast fundraising success
- Integration with broader fundraising ecosystems
PitchSense aims to empower founders by turning AI into a trusted co-pilot, blending efficiency and human insight seamlessly.
# Clone repository
git clone https://github.com/Avikalp-Karrahe/pitchsense.git
cd pitchsense
# Setup frontend
cd Front-end\ pitchsense
npm install
npm run dev
# Setup backend
cd ../server
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn server.main:app --reloadOpen http://localhost:8501 to interact.
- Frontend: Next.js, Tailwind CSS
- Backend: FastAPI, Python, GPT-4, Anthropic Claude
- Database: CSV, FAISS embeddings
- Infrastructure: Vercel, AWS Lambda
├── Front-end pitchsense/ # Next.js frontend application
│ ├── public/ # Static assets
│ ├── src/ # Source code
│ ├── match_api.py # Matching helper used by frontend
│ ├── eslint.config.mjs # ESLint configuration
│ ├── next.config.ts # Next.js configuration
│ ├── postcss.config.mjs # PostCSS configuration
│ ├── tailwind.config.ts # Tailwind CSS configuration
│ ├── tsconfig.json # TypeScript configuration
│ └── README.md # Frontend-specific README
├── server/ # FastAPI backend
│ ├── llm/ # LLM integration modules (OpenAI, Anthropic)
│ ├── routes/ # API routes (investor-match, pitch-gen, etc.)
│ ├── main.py # Backend entrypoint
│ └── __pycache__/ # Compiled Python cache
├── data files/ # CSV datasets for analysis
│ ├── VC_FundStage_Location_Sector.csv
│ ├── Startup Insights (2012–2021).csv
│ └── vc22.csv
├── Python scripts/ # Standalone utility logic
│ ├── agent_runner.py
│ ├── matching.py
│ ├── pitch&email.py
│ ├── confidence_scorer.py
│ ├── generator.py
│ ├── clarifier.py
│ ├── improver.py
│ └── llm_router.py
├── requirements.txt # Backend dependencies
├── .env # Environment variables (not committed)
└── README.md # This documentation
| Name | Role | GitHub | |
|---|---|---|---|
| Rachel Guo | Frontend Lead | GitHub | |
| Chaitanya Khot | UI/UX Lead | GitHub | |
| Yifei (Lexie) Li | Backend Lead | GitHub | |
| Avikalp Karrahe | AI Systems Lead | GitHub |
Distributed under the MIT License.
Special thanks to the organizers of AgentHacks 2025—including Dex, AfterQuery, and AGI House—for fostering a space where ambitious agentic ideas like PitchSense could be built and celebrated.
You made it all the way here! Thank you for your time and support 🙌