| title | BJJ AI Agents |
|---|---|
| short_description | An AI-enabled BJJ coaching app with specialized agents. |
| emoji | 🥋 |
| colorFrom | blue |
| colorTo | purple |
| sdk | docker |
| sdk_version | 5.35.0 |
| app_file | main.py |
| pinned | false |
NOTE: This originated from a talk I gave at the Kansas City AI Agents Conference on June 25, 2025. Here is the original notebook used for that talk.
Run app locally with: uv run uvicorn main:app --reload --host 0.0.0.0 --port 7861
A comprehensive Brazilian Jiu-Jitsu AI coaching system with multiple specialized agents for different aspects of BJJ training and competition.
- 🤖 Coach Agent: Interactive BJJ coaching with technique explanations
- 📋 Game Plan Agent: Tournament strategy generation with RAG enhancement
- 🎯 Smart Router: Intelligent query routing to appropriate agents
- 📊 Progress Tracking: Learning progress monitoring and storage
- 🗄️ Database Management: Comprehensive data storage and retrieval
- 🎲 Training Examples: Generation of diverse training scenarios
- Set your OpenAI API key in the environment variables
- Ask questions about BJJ techniques, strategies, or tournament planning
- Use the Smart Router for automatic agent selection
- Track your learning progress over time
Handles general BJJ questions, technique explanations, and training advice.
Creates personalized tournament strategies based on division, weight class, and opponent style.
Automatically routes queries to the most appropriate agent based on content analysis.
The system maintains comprehensive records of:
- Student information and progress
- Generated game plans
- Learning progress tracking
- Agent performance evaluations
ai_agent_conference/
│
├── src/
│ ├── config.py
│ ├── database.py
│ ├── llm_utils.py
│ ├── evaluation.py
│ ├── agents/
│ │ ├── coach_agent.py
│ │ ├── game_plan_agent.py
│ │ └── router_agent.py
│
├── prompts/
│ └── ...
├── requirements.txt
├── pyproject.toml
├── main.py
├── README.md
└── README_webapp.md