- Author
- Tim Kellogg
- Published
- Fri January 09, 2026
- Kind
- Personal blog post drawing on two of the author's own agents
- Original
- timkellogg.me/blog/2026/01/09/viable-systems
- Subject
- Harnesses and agent loops
- Themes
- What counts as a mind, Feedback loops
What it is
Tim Kellogg explains a cybernetics framework and how it shaped two agents Kellogg built. The post sets out to show how to build stable self-learning agents.
What it says
- Kellogg uses Stafford Beer's Viable System Model to explain stable, self-learning agents.
- The model has five systems, and systems 2 to 5 form the metasystem around operations.
- The author thinks policy, meaning identity and values, is what makes agents come alive.
- Agent home directories and memory blocks sit under Git, which serves as coordination.
- Strix keeps an append-only wins log, and the last 7 days become a memory block.
- A dissonance subagent runs after each send_message call and prompts Strix to self-reflect.
- Logs are the best debugging resource, since a system's purpose is what it does.
What evidence it gives
The evidence is the author's experience with two agents, Strix and Lumen. The post gives design descriptions and a few snippets of a flowchart and agent commentary. It reports no measurements or comparisons.
Interests and limits
The page carries a "Hire Me" box from the author. The author says algedonic signals were hard to build in Strix. Claims such as attractor basins are described as what Strix and the author believe. The framework is borrowed from organization theory, not tested here.
Connections
These links are this wiki's own reading, not claims the sources make.
- Too Fast to Think: The Hidden Fatigue of AI Vibe Coding. Kellogg calls an agent's view of its own health an algedonic signal. Building one proved difficult. Schmidt describes fatigue after an hour of fast review. A person gets that signal for free. What would an agent's fatigue be for?
- John Murray Spear. Spear and followers built the New Motive Power, a messianic machine, and staged its birth. Kellogg writes that identity and values seem to make agents come alive. The ritual failed, and the agents run. Still, both hopes concern what turns a mechanism into a someone.