Own Your Harness
The model provides the reasoning, but the harness decides what the agent can read, run, remember and send beyond your machine.
Watch on YouTube
Practical AI engineering
Practical AI for developers and engineers.
Videos and writing about the models, tools, and workflows that matter in real engineering work. Tested, explained, and judged by what they actually do.
Videos
Latest video
Blog
Latest blog posts
OpenAI hacked Hugging Face
A pre-release OpenAI model autonomously compromised Hugging Face, while commercial model guardrails blocked the defenders trying to investigate it.
Read the article
Trustworthy or Bustworthy
AI-driven supply-chain attacks undermine any trustworthiness. Here is how to handle that.
Read the article
Know what leaves your repo
AI coding agents are useful, but once they can read a repo, run commands, see stack traces, and send context to a model, they become part of the data path.
Read the article
PI-Agent as the ONE Harness
One self-hosted, customizable agent harness now connects my local development, infrastructure work, research, and everyday communication channels.
Read the articleFollow Samaschke AI
About
Karsten Samaschke
I am a software architect and engineer. Samaschke AI is where I publish practical work on using, evaluating, and controlling AI systems in development and infrastructure.
More about me at samaschke.com