Home

Welcome to the predikit Wiki

predikit is designed to bridge the gap between static Machine Learning models and dynamic LLM Agents. This wiki provides in-depth guides on how to handle complex data types, manage model registries, and build robust agentic workflows.

📖 Documentation Sections

1. Core Concepts
   • How ModelTool works under the hood.
   • The importance of Schema Mapping.
2. Agent Frameworks
   • Deep dives into LangGraph, AutoGen, and CrewAI.
3. Model Registries
   • Connecting to MLflow and Snowflake.
4. Advanced Features
   • Confidence-aware routing and Ensembles.
5. Production Best Practices
   • Error handling, logging, and performance.

---

Quick Links

• PyPI Project
• Issue Tracker