According to MIT NANDA, 95 percent of organisations are so far getting no return from generative AI (preliminary findings, July 2025). Gartner predicts that over 40 percent of agentic AI projects will be cancelled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls (forecast of 25 June 2025).
I read this as a hint not to decide on the technology first.
Three questions before approval
Where does the data stay? Who is liable? What happens to the roles in the team? As long as these questions are open, no tool helps.
Calculate first, then pilot
After that, I work out three to five use cases by benefit, effort and risk. One use case goes into live operation as a pilot, with a measured result. Breadth follows after that. At a growing technology company, we introduced HubSpot including its AI functionality. The platform is in daily use there today.
The first conversation takes 30 minutes and is free and without obligation. More at AI consulting for mid-sized companies.
Sources: MIT NANDA, “The GenAI Divide. State of AI in Business 2025”, preliminary findings, July 2025. Gartner, forecast of 25 June 2025.