Chatbot¶

About¶

The chatbot app connects Simmate to a large language model (LLM) using LangChain. It provides an agent with chemistry tools (e.g. looking up CAS numbers) and a table for storing chat history.

Installation¶

Add chatbot to the list of installed Simmate apps with:
```
simmate config add chatbot
```
Install the chatbot dependencies in the same environment as Simmate:
```
uv pip install "simmate[chatbot]"
```
Configure your LLM provider in ~/simmate/settings.yaml. The supported providers are OpenAI (via Azure) and Google-GenAI. Any other keys are passed directly to the LLM client.
```
chatbot:
    provider: "Google-GenAI"
    model: "gemini-2.5-flash-lite"
    api_key: a1b2c3d4....
    temperature: 0
```
Update your database to include custom tables from the app:
```
simmate database update
```

Basic Use¶

from simmate.apps.chatbot.agent import get_agent

agent = get_agent()
agent.invoke(
    {"messages": [{"role": "user", "content": "What is the CAS for aluminum oxide?"}]}
)