Building a Local AI Agent

Recently I was working on building a local AI agent Iris. An open-source project experiencing how to build a practical AI assistant that runs locally and can do more than just answer questions.

My focus is on the foundations of a capable agent: minimal planning, tool use, and breaking complex goals into manageable steps. The challenge is making these workflows reliable and useful, even when working with smaller local language models.

Iris is in early stages and I continue with making it an AI agent that are practical, transparent, and under your control. I would love to hear your feedback and ideas on how to make it better.

Explore Iris: https://github.com/m4ns0ur/iris