A self-hosted personal AI agent built on the Hermes model. Current work at Kardia Conseil.
Problem
A genuinely useful personal AI agent needs to run on infrastructure the team controls rather than a black-box hosted assistant, especially once it starts handling real day-to-day agentic tasks.
Solution
Building a personal AI agent on top of the Hermes model, giving it tools and memory to handle day-to-day agentic tasks directly instead of through a third-party hosted assistant.
Architecture
The Hermes model runs as the agent's reasoning core, wired to a tool-calling layer for its day-to-day tasks, with the agent's state and memory kept under direct control.
