Many companies are investing in AI without seeing much value from it, Mindstone chief executive Joshua Wöhle argues. His proposed fix is not another round of prompt tips: train people to use AI alongside their judgement, on problems they actually face at work.
Watch Desk analysis
What happened
In a Business Reporter article published by The Independent, Wöhle says fewer than 5 per cent of organisations are getting value from AI, based on Mindstone’s work training teams at large companies. The article also cites a 2024 BCG survey of 1,000 executives in 59 countries, which found 74 per cent of companies had yet to show tangible value from AI investments.
Wöhle’s prescription is to focus on augmentation rather than automation. He says Mindstone uses live demonstrations tailored to people’s roles to help them see how AI might fit into their work. He estimates that only about 10 per cent of the work is teaching people to use the tools, with the rest focused on changing how they think about working alongside them.
Why it matters
Those claims come from a company that sells enterprise AI training, and the article identifies Mindstone as a Business Reporter client. That commercial interest matters, particularly for the headline figure. But the practical argument is worth considering: a generic tool tutorial may not show a marketing team, analyst or other specialist where AI fits into their actual responsibilities.
Wöhle also draws a useful distinction. AI can make mistakes, so treating it as a way to remove people from a process may set teams up for disappointment. Using it to support people who can check and shape its output is a different proposition.
Our read
The article makes a case for role-specific practice, not proof that Mindstone’s approach works across organisations. Still, it points to a more useful test for AI training than counting completed courses: can staff use the tools on real tasks, recognise when the output is wrong, and explain what they contributed?
A live demo may help people picture the work. It will not, by itself, show whether the tool is reliable or the training improves results. That takes follow-up beyond the demo room, where the awkward questions about quality and value tend to live.
What to watch
- Whether organisations measure outcomes from role-specific AI training, rather than attendance or tool usage alone.
- Whether training teaches people to check AI output as well as produce it.
- Whether Mindstone or other providers publish evidence comparing these approaches with conventional tool training.
Discussion spark: Should companies judge AI training by how well staff can use it on real work and check the results, or by measurable productivity gains alone?
Sources and evidence
- Source update (2 November 2026, 17:45 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.