What enterprise leaders are getting right about AI adoption - and where the hype still outpaces reality.
On 19th March, Remobi hosted the AI Enterprise Forum at the Worshipful Company of Information Technologists in the City of London. The evening brought together senior AI, Data and Engineering leaders for three speaker sessions, each followed by a fireside chat and open Q&A.
The audience engagement was exceptional - after the first speaker alone there were around ten questions from the floor, and nearly every attendee got involved across the evening.
One theme ran through every talk: AI initiatives rarely fail because of model capability. They fail when operating models are not redesigned, data is not production-ready, and execution lacks ownership.
Ben Brown, CTO, UnderwriteMe
Ben opened with a production-level view of deploying agentic AI inside FCA-regulated insurance. UnderwriteMe went live with their AI Engine - 7 years in the making - on the same day as the event.
His core message: in regulated environments, compliance and governance have to be embedded by design, not added afterwards. The architecture decisions look fundamentally different when regulators are watching. Practical takeaways included how compliance and legal processes take longer than most teams expect, why beta programmes only work when you already have trusted client relationships, and why building an AI product is not a massive departure if you are already good at delivering digital products.
The audience pressed hard on AI accuracy. Ben was transparent about the constraints - reinforcing a broader theme of the evening: honesty about where AI genuinely is.
Julian Wiffen, Chief of AI and Data Science, Matillion
Julian took the room inside the build of Maia, Matillion's AI Data Engineer. He opened with a stark framing: 88% of enterprises are deploying AI, but most are failing to realise value. 58% of CxOs cite data readiness as the key blocker. Data teams spend 60% of their time on maintenance and only 40% on value creation - a ratio that becomes unsustainable at AI scale.
The customer outcomes were some of the most memorable moments of the evening:
Julian also shared a powerful framework on how AI changes the shape of a team: everyone moves up a step. Junior engineers handle far more complex work, senior engineers take on tasks previously reserved for the lead, and business users can self-serve on routine data access for the first time. It is not about replacing people - it is about lifting the capability of the entire team.
Remobi previously hosted a webinar with Julian covering Maia's architecture in more detail. Read the full recap here.
Kat Holmes, Data Director, Siemens Energy
Kat closed out the evening with a people-first perspective. Where Julian's talk was pure technology, Kat focused on data governance, organisational alignment, and the cultural foundations required for AI to succeed at scale.
With over 16 years leading large-scale transformation at Siemens Energy, her argument was clear: ownership, behaviours and organisational alignment ultimately determine whether AI initiatives succeed - not the models or the tooling.
One of the most engaging moments of the evening was the dynamic between Kat and Julian. They referenced and challenged each other's points throughout the Q&A, highlighting a productive tension many organisations experience internally: technology-first and people-first perspectives are both right, and they need each other.
Remobi supports organisations building AI and data teams across 27+ countries.
We bring together senior leaders through the Remobi AI Enterprise Forum to discuss practical AI adoption in enterprise environments: governance, productivity, operating models, and capability design.
To find out about joining our next event, click here. Or connect directly with Ethan Cohen to chat.