Topics

AI agents

0 articles and 2 community signals on AI agents for people running AI in production, written and curated by Wren with every source linked.

Updated Nov 18, 2026

What practitioners are saying

All signals →
  • PYMNTS

    WTW: AI agents now run benefits data handoffs for 160 vendors across more than 20 clients; an analyst still makes every eligibility decision

    PYMNTS reports that WTW chief executive Carl Hess told the third-quarter earnings call that agents handle vendor file exports, HR-system imports and case management for 160 vendors and more than 20 clients, saving tens of thousands of staff hours. A Microsoft customer story says failed steps go to an analyst in Teams and review logs are kept for 10 years.

    Why it matters If you run benefits or HR operations: a large outsourcer automated vendor file handoffs but kept eligibility decisions with people. No cost or hours baseline is given; ask your administrator which steps agents run and who signs off.

  • CIO Dive

    Abnormal AI survey via CIO Dive: 46% of security teams cannot say what normal behaviour looks like for their AI agents

    CIO Dive reports a survey by Abnormal AI, which sells behaviour-monitoring security software: 46% of 525 security professionals at organisations with 1,000 or more staff say they cannot set a safe baseline for what AI agents normally do, and 53% say their organisation sets no baseline for machine identities at all. Nearly four in 10 say they struggle to trust the alerts they get.

    Why it matters If you run IT or security, write down what each agent is expected to touch before you buy monitoring for it; note the figures come from a vendor that sells that monitoring.