tinychat

tinychat is a framework for building and deploying conversational AI systems. It provides three fundamental abstractions that allow engineers to conceptualize complex conversational flows and applications as simple message processing systems:

1. Message - Information quanta (the bits)
2. MessageProcessor - Information transformations (the channels)
3. CompositeProcessor - Information topology (the graph)

Everything else is derived. Databases, APIs, LLMs, CRMs are all processors, while conversations and agents (e. g., chains, graphs, etc.) are composite processors.

Quickstart

To get started, you can run our quickstart examples and get familiar with the framework. To start, create a virtual environment and install the dependencies.

python -m venv .venv
source .venv/bin/activate
uv install

Then, you can run the ping pong example with the following command:

make run

Documentation

For detailed documentation, please refer to our rules.