Auto LLM in HATZ: Intelligent AI Model Routing for Engineers

Introducing Auto LLM for model selection

The speaker introduces Auto LLM as a solution to the challenge of choosing from many AI models, explaining its core function as an intelligent routing system.

Auto LLM routes per message, not conversation

A crucial detail is highlighted: Auto LLM analyzes and routes each individual message to the best model, allowing for dynamic switching within a single chat.

Demonstration: MCP servers explanation

The speaker demonstrates Auto LLM by asking for an explanation of MCP servers, showing it routes to Miniax M2.5 and uses sub-agents like Claude Haiku 4.5 for efficiency.

Demonstration: Infographic creation

The system automatically switches to Gemini 3.1 Flash (Nano Banana 2) to create an infographic based on previous learning, showcasing its ability to handle visual tasks.

Demonstration: Word document generation

For a more complex task requiring code execution, Auto LLM routes to GLM5 to generate a formatted Word document, illustrating its advanced capabilities.

User control and manual override

The speaker explains that users retain control and can manually select models or switch back to Auto LLM at any time, ensuring flexibility.

Conclusion: Auto LLM's efficiency benefits

The video concludes by summarizing Auto LLM's value proposition: providing the right model for the right message automatically, enhancing user efficiency in Hats.