Add AI Pitch Coach for investor pitch coaching

[aziz-bouziri]

Pull Request Template for Agentic Coaching Applications

Title

Add AI Pitch Coach: Step-by-step coaching for startup founders to craft investor-ready pitches

Type of Change

• New Agentic Coaching Application
• Enhancement to Existing Application
• Documentation update
• Bug fix
• Refactoring
• Other (please specify):

Description of the Application

AI Pitch Coach is a warm, step-by-step guide for founders to craft an investor-ready pitch. The agent helps entrepreneurs overcome one of the most challenging aspects of fundraising: articulating their vision in a way that resonates with different investor audiences.
The Problem: Many founders struggle with pitching: distilling months of work, passion, and vision into a compelling narrative that convinces investors. Even great ideas can fail to secure funding due to poor pitch delivery.

The Solution: This application provides three distinct coaching modes powered by CrewAI agents that understand pitch structure, audience psychology, and persuasion techniques.

• Primary Features:

  • Coach Mode: Step-by-step guidance through 10 pitch stages (one-liner, problem, solution, UVP, target customer, market, business model, traction, moat, ask)
  • Refine Mode: Analyzes and rewrites existing pitch content with actionable improvements
  • Q&A Mode: Simulates investor questions and provides guidance on strong responses
  • Interactive Web UI: Real-time progress tracking, session persistence, and pitch export
  • Scoring System: Real-time clarity, persuasion, and confidence metrics
  • Session Management: SQLite persistence to continue iterating on pitches across sessions

• Tools Integrated:

  • OpenAI GPT-4o-mini (via CrewAI)
  • FastAPI (web framework and API)
  • SQLite (session persistence)
  • LangChain (agent orchestration)
  • Custom routing and scoring tools

• Expected User Interaction:

  • Users select a coaching mode (Coach, Refine, or Q&A)
  • Configure audience type (VC, Angel, Accelerator, Corporate) and funding stage
  • Chat with AI agents that provide personalized feedback and guidance
  • Iterate on responses to build a complete, polished pitch
  • Export final pitch as Markdown

Testing and Performance Validation

• Unit tests added and pass locally (test_api.py with mocked agent responses)
• Integration tests with the broader coaching platform
• Performance metrics: All API endpoints respond within acceptable timeframes (<30s for AI responses)
• Manual testing across all three modes with various pitch scenarios
• Live deployment tested on Render: https://ai-pitch-coach-0z0m.onrender.com
  Test Results:
• Health endpoint: 200 OK
• Coach mode: Successfully guides through pitch stages
• Refine mode: Provides actionable improvements and rewrites
• Q&A mode: Generates relevant investor questions with response guidance
• Session persistence: Sessions saved and retrievable across requests
• Web UI: All features functional with proper error handling

Impact on the System

This is a standalone agentic coaching application with minimal impact on existing systems.

• Dependencies:

  • fastapi>=0.110.0 - Web framework
  • uvicorn>=0.27.0 - ASGI server
  • crewai - Agent orchestration framework
  • langchain, langchain-openai, langchain-community - LLM integration
  • python-dotenv - Environment management
  • pyyaml - Configuration files
  • pydantic - Data validation
  • pytest - Testing framework

• Impact on Existing Agents:

  • No direct interaction with existing agents
  • Follows the same patterns as other coaching applications in the repository
  • Can be deployed independently via Docker or cloud platforms
  • Shares similar architecture patterns (FastAPI + AI agents) for consistency

Documentation and Usage Guide

• Documentation updated in the README.md
• New usage guide added for all three coaching modes
• Example use cases provided (Coach, Refine, Q&A mode examples)
• Deployment guide added (Local, Docker, Render)
• Screenshot guide provided (SCREENSHOT_GUIDE.md)
• API documentation available via Swagger UI at /docs

Checklist for Submission

• My code adheres to the project's coding and style guidelines
• I have thoroughly tested my agent with different use cases
• I have commented on my code to explain complex logic or integrations
• Documentation has been updated accordingly
• All tests pass locally without issues
• The submission includes no new unresolved issues or warnings
• I have reviewed the application's performance on edge cases

Related Issues and PRs

N/A - This is a new standalone coaching application

Screenshots and/or Video Demo

Sccreenshots found under screenshots folder
Live Demo: https://ai-pitch-coach-0z0m.onrender.com
API Docs: https://ai-pitch-coach-0z0m.onrender.com/docs

Reviewer Instructions

Please review the AI Pitch Coach application with focus on:

1. Agent Integration: Verify CrewAI agents are properly configured in src/pitch_coach/config/ (agents.yaml, tasks.yaml)
2. API Design: Check FastAPI endpoints in src/pitch_coach/api.py for proper request/response handling
3. User Experience: Test the live deployment or run locally to experience all three coaching modes
4. Code Quality: Review agent orchestration logic in src/pitch_coach/crew.py and tools in src/pitch_coach/tools/
5. Documentation: Ensure README provides clear setup and usage instructions

To test locally:

cd community_submissions/coach_pitch_coach
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env
# Add your OPENAI_API_KEY to .env
uvicorn src.pitch_coach.api:app --reload --port 8000

---

Additional Notes or Comments

This project was built to solve a real problem I face: learning to pitch effectively as someone aspiring to start a company. It combines my interest in AI agents with a practical use case that could help other founders struggling with the same challenge.

---

Submission Guidelines for AI Engineers

• Be sure to submit only fully tested, documented applications.
• Review the CONTRIBUTING.md for additional guidelines.