{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiem4txrggjnd73fbsp3r6rrx5odug4ilrfdlyoqsqj3nudc2prjle",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mnkfjxev5gq2"
  },
  "path": "/t/in-chat-recommendation-of-mcp-apps/1382367#post_4",
  "publishedAt": "2026-06-05T14:29:50.000Z",
  "site": "https://community.openai.com",
  "textContent": "apoo373:\n\n> For. e.g. If a Brand has an App (like Expedia) and the user asks Chatgpt to book a flight, would Chatgpt surface the MCP App, Flight listings on the internet, some flight inventory from ACP or Ads or flights or maybe Chatgpt would recommend going to a platform and booking the flight etc etc)\n\nIn general, any answer I have here will not be correct for very long; we run a lot of experiments here for the classic measure like engagement/retention but also to make sure users actually like the results (“win rate” in industry terms). Things like memory and personalization add another layer of challenge here. (And I won’t ignore or hide, of course, that we have a pretty strong incentive not to share too much around the algorithm to avoid it getting hacked.)\n\nWhat I can say is that we’re going to start simple and iterate over time.\n\nI know there’s a lot of information and misinformation out there about what ChatGPT prioritizes; my recommendation is to be your own source of truth and to try things out yourself and with friends/colleagues who maybe have different preferences and ways of using ChatGPT.",
  "title": "In Chat Recommendation of MCP Apps"
}