Skip to content
Boxed Online, home
All build notes

5 min readAI leadership

Why most AI pilots stall, and the three things that fix it

95% of organisations report no measurable return from generative AI. The 5% that do share three habits, and none of them is choosing a better model.

By Gary Kellett, founder of Boxed Online

MIT's 2025 study of generative AI in business found that 95% of organisations saw no measurable return. RAND found the most common root cause of AI project failure is simpler still: people misunderstood the problem they were trying to solve.

In our experience the pilots that stall share a pattern. They start with a tool rather than a process. They are judged on a demo rather than a number. And nobody owns them once the workshop ends.

The fix is unglamorous. Start with one process that runs every week and costs real hours. Agree the before-numbers in writing. Build it on real data, not a sample. Then give it an owner who is measured on the result.

That's exactly why we offer the first automation free. It forces both sides to pick a process that matters, measure it honestly, and prove it in the real world before anyone signs anything.

Next note

"There must be a better way": how an architect looks at a small business

Enterprise architecture sounds like something for FTSE companies. The same thinking, sized properly, saves small businesses from buying the wrong system.

Newsletter

Get the next note by email

Built this month

One email a month: what we built, what it was quoted at, and what it actually took. No fluff.

Tell us what's stuck.

In 30 minutes you'll know whether to fix it, buy it, build it or leave it. No pitch deck.

We reply within 2 business days.