{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiangoiquigr3jha7lw7bljqxbqhtripfcaa5pmfhpg4nuvfiab7we",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mlubb32dqco2"
  },
  "path": "/t/what-are-the-biggest-challenges-in-multi-agent-system-development/175760#post_9",
  "publishedAt": "2026-05-15T01:14:23.000Z",
  "site": "https://discuss.huggingface.co",
  "tags": [
    "https://medium.com/accredian/traditional-vs-late-chunking-ed1a70022ff4"
  ],
  "textContent": "I think one of the biggest issues is that chunking + embeddings break the original document structure.\n\nRAG can retrieve relevant slices pretty well, but it often misses the overall context and relationships between parts of the data.\n\nFeels like semantic similarity alone is not enough for complex reasoning.\n\nCurious what direction people think makes more sense:\n\n  * graph-structured documents?\n\n  * compressed representations?\n\n  * late chunking https://medium.com/accredian/traditional-vs-late-chunking-ed1a70022ff4\n\n\n\n\nOr something else entirely?",
  "title": "What are the biggest challenges in multi-agent system development?"
}