Always-On Agents: A Survey of Persistent Memory, State, and Governance in LLM Agents
Tianyu Ding, Aditya Nannapaneni, Bingfan Liu, Ling Zhang · 2026
Read Always-On Agents: A Survey of Persistent Memory, State, and Governance in LLM Agents by Tianyu Ding, Aditya Nannapaneni, Bingfan Liu, Ling Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-30306.txt.
Introduction
Tianyu Ding, Aditya Nannapaneni, Bingfan Liu, and Ling Zhang survey AI agents that never switch off, focusing on the state they accumulate. Always-on agents remember across sessions, unlike one-shot systems. Their durable state includes far more than memories. Task ledgers, permissions, credentials, commitments, and audit records all count. The survey reads the literature through six axes.
Each state item is scored for authority, scope, and mutability. Provenance, recoverability, and actionability complete the six axes. A lifecycle follows state from writing through validation and retrieval. Forgetting, auditing, and rollback close the loop. The coded corpus spans four hundred thirty-five works. Most work focuses on storing and fetching state, the authors find. Governing, recovering, or giving up state gets much less attention. Their pilot evaluation protocol scores mutation and recovery duties directly.
Catalog record
| CREATOR | Tianyu Ding, Aditya Nannapaneni, Bingfan Liu, Ling Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 400 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.30306