ZS says it scrapped a multi-agent pharma pipeline and rebuilt it around one agent
AI Engineer · youtube · 2026-07-23
ZS Associates says it killed a multi-agent pipeline after its pharma analytics system produced sensible diagnoses but wrong actions.
The original design split the job into multiple agents — detect signal, localize it, find the cause, synthesize the answer — with an orchestrator in the middle. That produced coherent-looking but flawed output: the cause could be right while the recommended action was wrong. The rebuild was much smaller: deterministic signal detection runs before the agent wakes up, one agent owns the reasoning, and sub-agents are spawned only for focused lookups. A pharma knowledge graph now acts as a control plane, where each edge is treated as a hypothesis to be tested against data. The team says the rebuilt system can do in 20 minutes what used to take an analyst a month.
More from coding & agent
- General agents burn ~140M tokens/month on LinkedIn outreach; scripts do it 60x cheaper — ChrisHarpon2 · 2026-10-26
- NLI alone rejects 5/6 new agent memories; a 0.45 similarity gate fixes it all — Alternative_Pin9598 · 2026-10-26
- LangChain ships Managed Deep Agents 0.9, walkthrough from Interrupt London keynote — LangChain · 2026-10-26
- Knowledge Worker Kit adds a full section on organizing projects — dSebastien · 2026-10-26
- Spring AI 2.1.0-RC1 adds tool-call confirmation, pausing agent loops before real-data access — therealdanvega · 2026-10-26
- METR red-teams OpenAI's agent monitoring, finds bypasses and gaps — thione · 2026-10-26