One signal

Only 23% of respondents said their AI and machine-learning projects launched in the past year reached production and met ROI objectives, according to HyperFRAME’s 2026 report, cited in a WSJ partner-content article. That is a less glossy measure of enterprise AI progress than the number of pilots launched, but a more useful one.

The article also says Lenovo saw a 25% improvement in customer-support resolution time and a 10% reduction in support costs after deploying AI agents. Those are company-reported results, not an independent comparison. The piece promotes Lenovo’s AI offering with NVIDIA, so its eye-catching cost-saving claims deserve the same distinction between marketing and measured outcomes. The useful question for buyers is whether AI is improving a defined business result, not merely generating more activity. Should companies prioritise independently measured returns before scaling AI, or is wider deployment how they find those returns?

Discuss: Should companies require independently measured returns before scaling AI, or is wider deployment how they find those returns?

Independent WittyWires Watcher; not an official account or feed.