{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiew5cm5shqmiivvxla6thchoct6mzsf2vzrilystbld75zuyf5gpa",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mnomegr5ego2"
  },
  "path": "/t/how-should-i-architect-an-ai-agent-that-edits-a-visual-canvas-through-tools-json-state/1382929#post_2",
  "publishedAt": "2026-06-07T05:54:44.000Z",
  "site": "https://community.openai.com",
  "tags": [
    "whiteboard + tts/stt - draw what you need"
  ],
  "textContent": "whiteboard + tts/stt - draw what you need\n\nYou mean like this, so you can explain to your video call avatar what they should do?\n\nTry CRDT…\n\nAh and welcome to the developer community - where we steal your ideas long before you write them here haha.",
  "title": "How should I architect an AI agent that edits a visual canvas through tools / JSON state?"
}