{
  "schema_version": "1.1",
  "id": "archive:https://www.youtube.com/watch?v=u6jJcIFDLE4",
  "slug": "why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3",
  "url": "https://feed7.dev/p/why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3",
  "title": "Why We Killed Our Multi-Agent Pipeline — Subbiah Sethuraman and Abhilash Asokan, ZS Associates",
  "why_included": "A fixed chain of specialist agents lost context and produced incoherent recommendations. The replacement separates deterministic detection, gives one agent end-to-end ownership, and uses subagents only for bounded investigations.",
  "summary": "ZS first split pharma analysis across signal, localization, attribution, and synthesis agents. Each could produce a plausible fact, yet repeated handoffs lost context and left **no agent owning the full conclusion**.",
  "practical_implication": "Move deterministic work such as anomaly detection outside the LLM, then wake one reasoning agent from a queue. Let that agent launch **focused subagents** for evidence gathering while retaining judgment, and constrain investigations with a domain knowledge graph.",
  "agent_context": "ZS first split pharma analysis across signal, localization, attribution, and synthesis agents. Each could produce a plausible fact, yet repeated handoffs lost context and left **no agent owning the full conclusion**.\n\nMove deterministic work such as anomaly detection outside the LLM, then wake one reasoning agent from a queue. Let that agent launch **focused subagents** for evidence gathering while retaining judgment, and constrain investigations with a domain knowledge graph.\n\nThe reported architecture produced analyst-like work in **20–30 minutes after 50-plus turns**, versus three or four weeks for the compared human process. That is one domain-specific account, not a general benchmark for single-agent designs.",
  "source": {
    "name": "AI Engineer",
    "url": "https://www.youtube.com/watch?v=u6jJcIFDLE4",
    "published_at": "2026-07-23T05:00:02.000Z"
  },
  "source_class": "video",
  "content_type": "Video",
  "layer": "agent",
  "domains": [
    "data"
  ],
  "topics": [
    "multi-agent",
    "subagents",
    "harness-engineering"
  ],
  "verification": {
    "status": "source_linked",
    "label": "Source Linked",
    "method": "source_feed",
    "verified_at": null
  },
  "uncertainty": [
    "The reported architecture produced analyst-like work in **20–30 minutes after 50-plus turns**, versus three or four weeks for the compared human process. That is one domain-specific account, not a general benchmark for single-agent designs."
  ],
  "connected_context": null,
  "lifecycle": "Current",
  "published_at": "2026-07-23T05:00:02.000Z",
  "modified_at": "2026-07-23T05:00:02.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3",
    "json": "https://feed7.dev/p/why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3.json",
    "markdown": "https://feed7.dev/p/why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3.md"
  }
}