When AI pilots succeed on a measure customers do not care about
Egiziago Cioffi is CEO of SYNSPHERE ITALIA, which builds and operates AI and cloud systems for Italian mid-market firms, European insurers and banks. The conversation explores why the company is accountable for keeping systems running, what it has handed to AI, and how a pilot can succeed on a measure customers do not value.
Judge AI pilots by outcomes that matter beyond the project team, not by internal measures alone. Automating support can make sense, while decisions about who gets access still need human control.
What’s the quick origin story of your business, and what makes what you do genuinely different from the other options your customers are looking at?
Fifteen years at IBM, then watching the same mid-market firms get sold enterprise-shaped projects they could not run once the consultants left. The differentiator is framed as a design constraint rather than a service model: most suppliers are paid to deliver, we are paid to keep it running, so a clever decision that turns expensive in year two is one you pay for yourself. Saying no follows from that.
Where does most of your new business come from today, and where do you wish it came from?
Referrals, named as both the comfortable and the fragile answer, because the network grows only as fast as the people already in it. Then the press work, stated openly. And the wish: that clients called eighteen months earlier, before an auditor has written a date in the diary.
What’s one thing you’ve actually handed over to AI in the last year, and one thing you tried it on and went back to doing the human way?
First-line support with the 60%, the three roles gone as a function, and the 5.12% reopen rate chosen because the customer controls it. Taken back: deciding who may see what, told through the permissions bug. Closes on we did not take the task back, we took back the right to decide who sees what.
Have you noticed a change in how customers find you or what they already know before they reach out? Is anyone arriving through AI assistants like ChatGPT yet?
The lever is that what got syndicated to fourteen sites in seven languages was a failure, not a success, plus your own argument that generative engines cite witnesses. Closes on the page that brings us work is not on our website.
If you were advising someone in your industry on all of this for the next 12 months, what would you tell them to actually do, and what would you tell them to ignore?
Three things to do, including the warning about the satisfaction survey you had built before noticing what it rewarded, and three to ignore. Closing line, which is the headline material: most AI pilots do not fail, they succeed on a measure nobody outside the project cares about, and then quietly stop.
Thank you to Egiziago Cioffi and the team at SYNSPHERE ITALIA for sharing what actually worked with Leaders Perception readers.
Want to share your playbook?
Leaders Perception publishes founder and operator interviews every week. A few questions, about five minutes, no calls. It is free and everyone we accept gets published.