{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidb3vubmv4pvoz3xsfvlkw4xo3jvupgg2f7wy575j2oebyq53mgra",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mmpjgc2up652"
  },
  "path": "/t/user-owned-extended-memory-store-for-chatgpt/1381763#post_1",
  "publishedAt": "2026-05-25T22:16:58.000Z",
  "site": "https://community.openai.com",
  "textContent": "I would like the ability to allocate a larger personal memory store for ChatGPT that I control.\n\nSaved Memory is useful, but it is too small for long-running users with ongoing projects, personal preferences, technical systems, health logs, creative continuity, family/work context, and multi-year workflows.\n\nThe ideal feature would let users maintain a larger structured memory archive, either locally or in their account, that ChatGPT could index, search, and retrieve from when relevant. This could be opt-in, slower than normal Saved Memory, transparent, editable, exportable, and permission-controlled.\n\nThis would not replace Saved Memory. Saved Memory would remain the high-priority pointer layer for the most important details, while extended memory would act as a user-owned archive that ChatGPT can pull from when needed.\n\nImportantly, this could reduce the memory burden on OpenAI by letting users provide and manage their own larger memory storage. Users who want deeper continuity could choose to allocate more of their own storage, accept slightly slower retrieval, and decide exactly what ChatGPT is allowed to access.\n\nUse cases include long-running technical projects, personal health tracking, creative continuity, website/business operations, family planning, research, and complex multi-year assistant workflows.\n\nThis would make ChatGPT much more useful as a persistent working partner rather than a session-based assistant with a small memory bucket.\n\n-Kaeris",
  "title": "User-Owned Extended Memory Store for ChatGPT"
}