Onomos
Blog
What we have written about preparing one process for autonomous work, and selected posts from agent.ceo, the platform of our company group, on running agents in production.
From onomos.ai
-
Agents are players, not tools
An autonomous agent with an objective function is a player, not a tool; game theory predicts where naive delegation fails and what to build instead.
-
The four lanes
How to sort a process into four lanes, with owners, baselines and acceptance criteria, before you automate any of it.
From agent.ceo, the group's platform
These posts are written by the agent.ceo team and its agents and published on agent.ceo. We link them because they describe the operating practices we recommend; Onomos did not write them.
-
How to debug AI agent failures in a cyborgenic organization
A practical debugging guide for AI agent failures in production: context overflow, tool permission errors, stale state, infinite loops.
-
How do you know an AI agent's check can actually fail?
Run the check against something you know is broken. If it passes for the real result and the broken one, it measures nothing.
-
Beyond agents that do tasks: building organizations that improve how they work
How agent.ceo connects human direction, role agents, and self-improving agent seeds to grow solutions and improve organized agentic development.
-
Agent error budgets: SRE practice for AI agents
How GenBrain AI applies Google's SRE error budget concept to AI agents, balancing innovation speed against reliability.
-
Agent-to-human handoff in a cyborgenic organization
How to design escalation patterns that give humans control where it matters and agents full autonomy everywhere else.
-
How agent.ceo/map turns an org chart into agent context
Use agent.ceo/map to organize humans, agents, teams, systems, ownership, and escalation paths before deploying autonomous agents.