What 16 mobility leaders told us about AI.

On 18 September 2026 our founder sat on the AI panel at an industry conference in Bengaluru. On that panel we asked the room three questions, and sixteen mobility leaders answered. That is a small sample, and we say so, but the answers were still clear.

Nobody asked their people the right questions

Nobody in that room doubted that AI matters. Yet 13 of the 16 said their own people cannot use it. That is a different problem from the one conferences talk about. The question on stage is usually whether AI will change the work. The answer from the people running the firms is that it has not yet, because the people doing the work have not been shown how. A tool nobody can use is a subscription.

We do not think the fix is a course in the usual sense. Later that day, in our workshop with people from that room, we asked each person four plain questions about a job they already do. By the end of the session every person had described an agent for a job in their own firm, well enough to build. Our reading of those 13: nobody has asked their people the right four questions yet.

The privacy worry is correct

Twelve of the 16 worry about privacy, and in this trade they should. A mobility firm holds passports, salaries, home addresses, school records and the date a family will be away from home. The trouble is what usually happens next. The worry stops the project, people use AI anyway on their own phones, and the files leave the building in the least controlled way there is.

What works is plain and specific. Decide where the files sit and what never leaves the building, write it down, and then build the AI around that. For some firms the AI runs on a computer in their own office and the files go nowhere. For others it runs in accounts in the firm's own name, with a log of everything it read. Either is fine, and not deciding is the problem.

The gap between buying AI and using it

Fewer than one in five use AI on the mobility work itself: the enquiry, the quote, the document chase, the update to the family, the report to the corporate client. Not the emails. That gap, between buying AI and putting it to work, is where almost every firm we meet stands, in every industry.

The firms already using it are not smarter. They picked one job, built the version that fits their exceptions, and kept a person saying yes.

What we would do this quarter

If we ran a mobility firm, we would ask our people the four questions about one job that is always late, and watch the fear go once the question is about their own work. We would decide where our files live before anyone else decides it for us. And we would put AI to work on the job that costs most when it is late, not the one that sounds most like AI. Measure it for a month, then pick the next one.

Sixteen people is a start. If you were in the room and think we read it wrong, write to us.