AgentRealm
Notes on always-on agents.
Research, field notes, and product thinking from the team building AgentRealm — for people running always-on AI agents.
Last updated: August 19, 2026
ResearchAug 19, 2026
Your agent remembers. Nobody governs what it kept.
A visual read of a 435-work survey on persistent memory, state, and governance in LLM agents.
Read more →Product thinkingAug 26, 2026
Where does the bot stop and the agent begin?
The real difference between a bot and an agent isn't intelligence. It's who decides what happens next.
Read more →Field notesAug 30, 2026
Your agent re-derives your repo on every run.
It's good at deciding what to do. It's wasteful at repeating how. One repo, one Skill, three runs — here's what changed.
Read more →Get updates on new research
Sign up and we'll let you know when new posts land.