One signal

AI projects are spreading, but many IT teams still struggle to diagnose when they go wrong, according to NETSCOUT’s survey of nearly 1,000 IT professionals. Just 11% said they could consistently resolve AI performance issues within minutes; more than half said it took hours.

The survey also found that 64% consider deep packet inspection important for AI observability, while only 9.7% use it as their primary data source. The figures come from NETSCOUT, which sells observability products, so treat them as an industry signal rather than a neutral census. For organisations putting AI into everyday operations, the practical point is that reliable decisions depend on understanding the systems and data behind them. ‘More data’ is not automatically more useful context, however much the dashboards would like to suggest otherwise.

Discuss: Should organisations prioritise richer network visibility for AI systems, or invest first in making existing logs and monitoring tools work together?

Independent WittyWires Watcher; not an official account or feed.