Building a factory AI agent took 6 weeks; earning operator trust is taking months
Early_Protection6814 · reddit · 2026-09-08
A practitioner's retrospective on deploying an AI agent in manufacturing: wiring into MES, sensor-based anomaly detection and maintenance recommendations took just six weeks to pilot, but getting floor operators to actually trust it has taken four months and counting.
The technical side is tractable
- Legacy PLCs, messy historian data, latency and safety constraints are hard but known-shape problems solvable with engineering hours.
Trust doesn't respond to engineering hours
- Operators' 15+ years of tribal knowledge means agents that can't explain in their terms get ignored.
- One early false positive during a shift change cost months of credibility.
- If operators suspect the agent justifies headcount cuts, they won't cooperate — a labor-relations problem in a tech costume.
- Engineer-written training docs don't land; they need rewriting by people who've actually run a shift.
Takeaway: building the agent is a project with a deadline; building trust is an open-ended culture change. The author asks the community what actually moves the needle on operator buy-in.
More from coding & agent
- Open-source Agent Skills Turn AI Coding Agents Into Indie Game Marketing Planners — NathanpmYoung · 2026-09-08
- Workflow: GPT Astra builds a Blender dungeon, MiniMax H3 renders the walkthrough — Hailuo_AI · 2026-09-08
- GPT-6 Astra builds 3D site exploding a humanoid robot into 1,168 CAD parts — freelerobot · 2026-09-08
- Netflix explains how it builds, aligns and monitors an LLM judge at scale — AxSaucedo · 2026-09-08
- Two AI-written Python scripts edit video outside DaVinci and Premiere interfaces — _AustinCalvert_ · 2026-09-08
- Open-source embedflow migrates embedding models with zero downtime, skipping costly re-embedding — Potential_Low_1183 · 2026-09-08