Deployment
P&G cuts warehouse shift planning from 8 hours to 20 minutes
Daily workforce planning at Procter & Gamble's manufacturing campuses used to take eight hours per shift, and the plan usually broke within ten minutes as equipment failed or workers called in sick. IIoT World details the warehouse AI that cut it to 20 minutes.
A plan that breaks within minutes is a good AI target; time your own shift-planning effort before scoping a pilot.
Deployment
Teva evaluated 300 technologies for its 30 factories
Teva Pharmaceuticals evaluated about 300 technologies in one year while building an open innovation platform across its 30-factory network. Operations are around 70% of its cost structure, Maya Schushan-Orgad told the Industrial AI Summit 2026.
At 30 sites, selection discipline is the work; set evaluation criteria before you start collecting pilots.
Deployment
Why AI pilots stall before production
A $20,000 AI pilot runs on a handed-over spreadsheet, one or two APIs and a controlled scope. Production adds cybersecurity requirements, legacy equipment on different software versions, unplanned downtime and a workforce that may resist, per the Industrial AI Summit 2026.
Scope the pilot against production constraints (security, legacy versions, downtime) from day one, not after it works.
Agents
Is agentic AI an OT insider threat?
Agentic AI in operational technology acts with its own authority at machine speed, where a wrong decision moves a valve or opens a breaker. Black & Veatch and GrayMatter speakers at the Industrial AI Summit 2026 examined whether it should be treated like an insider threat.
Treat any agent with write access to OT as a privileged user: scope its authority and log its actions.
Workforce
Opinion: what happens when the robot is wrong?
Gleb Tsipursky argues an industrial robot can hit its cycle-time target and still fail the key workplace test: what happens when an experienced operator believes the machine is wrong. He raises it as manufacturers accelerate automation.
Write the operator-override rule before deployment: decide who wins when an experienced operator disagrees with the robot.