Konen: human-centered orchestration of AI agents

The Swiss army knife for agentic IREs

Konen is pronounced /ˈkoʊnɛn/. Its name comes from the star graph $K_{1,n}$, reflecting Konen's human-centered approach to agentic interaction. The human occupies the central node and orchestrates a collection of AI agents represented by the leaf nodes.

Konen emphasizes human-directed deliberation. A user decides which agents participate and how they engage with each other's contributions, while maintaining a shared conversation history. Although similar capabilities exist in other multi-model tools, Konen offers a simple, focused approach suited to iterative exploration of ideas, critical analysis, and research. Thus, Konen is suited to use as an integrated research environment (IRE) for coordinating interactions across AI agents and model providers.

Installation

Konen requires Python 3.13, 3.14, or 3.15.

pip

Install Konen from PyPI:

pip install konen

uv

To install the CLI in an isolated environment:

uv tool install konen

Run konen after installing it. For a source checkout and development setup, see DEVELOPMENT.md.

Documentation

Read the documentation, starting with the first walkthrough example.

License

Konen is distributed under the MIT License. See LICENSE.