--- schema: 1 type: video title: >- Why We Killed Our Multi-Agent Pipeline — Subbiah Sethuraman and Abhilash Asokan, ZS Associates description: >- Their first pharma analytics system mimicked a human analyst: one agent to detect a signal, one to localize it, one to find the cause, one to synthesize, all wired to an orchestrator. It produced answers like this: prescriptions dropped 18% in a territory because a payer moved the drug to a worse tier, so send more sales reps. The cause was right and the action was wrong, because no single agent owned the whole picture. So Subbiah Sethuraman's team at ZS killed the multi agent pipeline. Instead of redesigning the topology, they opened an empty directory, gave Claude Code bash and the database, and watched what it actually did. The rebuild came out smaller, not bigger. Signal detection moved into a deterministic pipeline that runs before the agent wakes up, so the agent investigates rather than guesses. A single agent owns the reasoning and spawns sub agents only when a focused lookup needs one. A pharma knowledge graph acts as a control plane, not a lookup table: every edge is a hypothesis the agent tests against the data, which bounds the search. The result does in 20 minutes what an analyst did in a month. Speaker info: - https://www.linkedin.com/in/subbiahsethuraman/ - https://subbiah-sethuraman.medium.com/ Timestamps: 0:00 - Pharma commercial analytics and the analyst's four steps 2:33 - V1: an agent for every step 3:26 - Why the output was incoherent 4:32 - Why it failed: signals, handoffs, and missing domain 5:57 - The rebuild: watching Claude Code in an empty directory 7:01 - Deterministic signal detection before the agent 8:05 - Consolidating to a single agent 9:22 - The knowledge graph as a control plane 11:04 - Every edge a hypothesis, and the result --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/u6jJcIFDLE4-why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-asokan-zs-associates Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- canonical: >- https://liquidgold.cc/channel/aidotengineer/explore/video/why-we-killed-our-multi-agent-pipeline-subbiah-seth-8570d2d7eed3 markdown: >- https://liquidgold.cc/channel/aidotengineer/video/why-we-killed-our-multi-agent-pipeline-subbiah-seth-8570d2d7eed3.md llms: null parent: 'https://liquidgold.cc/channel/aidotengineer.md' channel: aidotengineer updated: '2026-10-24T06:41:39.493Z' language: null children: [] videoId: u6jJcIFDLE4 durationSeconds: 900 publishedAt: '2026-07-23T05:00:02.000Z' searchChunks: - ordinal: 0 text: null startSeconds: 23 endSeconds: 155 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=23' heldReason: rights - ordinal: 1 text: null startSeconds: 23 endSeconds: 155 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=23' heldReason: rights - ordinal: 2 text: null startSeconds: 165 endSeconds: 209 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=165' heldReason: rights - ordinal: 3 text: null startSeconds: 220 endSeconds: 276 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=220' heldReason: rights - ordinal: 4 text: null startSeconds: 288 endSeconds: 366 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=288' heldReason: rights - ordinal: 5 text: null startSeconds: 288 endSeconds: 366 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=288' heldReason: rights - ordinal: 6 text: null startSeconds: 288 endSeconds: 366 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=288' heldReason: rights - ordinal: 7 text: null startSeconds: 378 endSeconds: 421 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=378' heldReason: rights - ordinal: 8 text: null startSeconds: 431 endSeconds: 496 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=431' heldReason: rights - ordinal: 9 text: null startSeconds: 431 endSeconds: 496 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=431' heldReason: rights - ordinal: 10 text: null startSeconds: 507 endSeconds: 570 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=507' heldReason: rights - ordinal: 11 text: null startSeconds: 582 endSeconds: 673 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=582' heldReason: rights - ordinal: 12 text: null startSeconds: 582 endSeconds: 673 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=582' heldReason: rights - ordinal: 13 text: null startSeconds: 683 endSeconds: 900 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=683' heldReason: rights - ordinal: 14 text: null startSeconds: 683 endSeconds: 900 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=683' heldReason: rights - ordinal: 15 text: null startSeconds: 683 endSeconds: 900 watch: 'https://liquidgold.cc/watch/u6jJcIFDLE4?t=683' heldReason: rights --- # Why We Killed Our Multi-Agent Pipeline — Subbiah Sethuraman and Abhilash Asokan, ZS Associates Their first pharma analytics system mimicked a human analyst: one agent to detect a signal, one to localize it, one to find the cause, one to synthesize, all wired to an orchestrator. It produced answers like this: prescriptions dropped 18% in a territory because a payer moved the drug to a worse tier, so send more sales reps. The cause was right and the action was wrong, because no single agent owned the whole picture. So Subbiah Sethuraman's team at ZS killed the multi agent pipeline. Instead of redesigning the topology, they opened an empty directory, gave Claude Code bash and the database, and watched what it actually did. The rebuild came out smaller, not bigger. Signal detection moved into a deterministic pipeline that runs before the agent wakes up, so the agent investigates rather than guesses. A single agent owns the reasoning and spawns sub agents only when a focused lookup needs one. A pharma knowledge graph acts as a control plane, not a lookup table: every edge is a hypothesis the agent tests against the data, which bounds the search. The result does in 20 minutes what an analyst did in a month. Speaker info: - https://www.linkedin.com/in/subbiahsethuraman/ - https://subbiah-sethuraman.medium.com/ Timestamps: 0:00 - Pharma commercial analytics and the analyst's four steps 2:33 - V1: an agent for every step 3:26 - Why the output was incoherent 4:32 - Why it failed: signals, handoffs, and missing domain 5:57 - The rebuild: watching Claude Code in an empty directory 7:01 - Deterministic signal detection before the agent 8:05 - Consolidating to a single agent 9:22 - The knowledge graph as a control plane 11:04 - Every edge a hypothesis, and the result --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/u6jJcIFDLE4-why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-asokan-zs-associates Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- Canonical: https://liquidgold.cc/channel/aidotengineer/explore/video/why-we-killed-our-multi-agent-pipeline-subbiah-seth-8570d2d7eed3 · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md) > Search chunk text is held until public-source provenance is verified. ## Document chunks > Missing: no stored document chunks are available.