Attending FinovateEurope 2026 in London, one thing was immediately clear: AI has moved from being a differentiator to being the default. The agenda, the demo floor, and the conversations in the room were all saturated with it. Almost every startup exhibiting had an “AI-powered” angle, which, visually at least, made standing out harder than ever. When everyone is under the same spotlight, it becomes difficult to see who is genuinely doing something different.
If Finovate has always been about product, this year’s edition stayed true to that. But the
real value wasn’t confined to the stage. It was found in the informal spaces: the impromptu
meeting zones, the tables at the back of the room, the lunch lines and the coffee queues.
That’s where many of the most useful conversations seemed to happen. In a slightly ironic
twist, the removal of charging points this year turned the search for the odd available plug
socket into an unlikely networking hub. Few things create fintech bonding quite like a dying
battery and a shared extension lead.
For companies serious about building with AI in mind, FinovateEurope 2026 was a useful
touchpoint. But it also underlined a more uncomfortable truth: not everyone is ready to
deploy at scale. The industry is no longer debating whether AI matters. It is grappling with
how to implement it in environments that are complex, regulated, and deeply constrained by
legacy infrastructure.
That tension sat at the heart of one of the most grounded sessions on the agenda: the
Power Panel, chaired by Theodora (Theo) Lau, which set out to get “beyond the hype” and
into how financial institutions can actually make or save money with AI. The panel brought
together Norman Tambach (Mashreq), Mei Lim (Anthemis), and Doruk Mutlu (Evam), and
focused less on futuristic promises and more on operational reality.
A recurring theme was the gap between promising ideas and production-ready use cases.
Asked to sum up AI’s near-term impact in one word, Doruk Mutlu chose “Amplified”, a
framing that resonated across the panel. AI, in this context, isn’t so much replacing teams as
stretching what already exists: skills, processes, and expectations, often in uncomfortable
ways.
Mei Lim offered one of the most pragmatic perspectives of the day, particularly from the
investor side. She described how many AI startups hit the same commercial barrier:
“typically around a very long sales cycle.” In practice, that means unclear and slow approval
processes, the need for multiple internal sponsors across business and compliance, lengthy
onboarding even after a deal is signed, and the risk that a key internal champion leaves
halfway through. In regulated environments, that cycle can easily stretch to 12–18 months, a
timeline that can quietly kill early-stage companies long before the technology itself fails.
The discussion moved quickly from diagnosis to mitigation. Panellists emphasised the
importance of mapping stakeholders early, preparing compliance and regulatory
documentation upfront, designing phased pilots with clear success metrics, and reducing
integration friction wherever possible. There was also a strong case made for ecosystem
approaches, such as shared engineering and affordable compute platforms like CommonAI,
launched by Anthemis and Cambridge AI Venture Partners, to stop smaller companies being
structurally disadvantaged by long enterprise sales cycles and heavy internal requirements.
If the panel was about operational realism, the keynote from Alpesh Doshi, Managing
Partner at Redcliffe Capital, was about structural readiness. His session, “AI First Banking –
Why Agentic AI is Truly a New Frontier in Banking,” argued that banks cannot unlock the full
value of AI without first getting serious about how they store, govern, and process data. In
highly regulated environments, Star Trek fans will recognise this as the difference between
moving at “warp speed” and “light speed”: the models may be powerful, but without the right
data foundations, they simply cannot perform.
Doshi framed the broader shift as a move from “augmented banking to autonomous finance,”
where agentic AI systems don’t just assist humans but plan, reason, and execute complex
workflows end-to-end. That includes everything from reporting and compliance to hyper-
personalised services. But he was clear about the barriers: legacy stacks, poor data quality,
security concerns, regulatory constraints, and the need to rewire how organisations think
about work. The message was not that these problems are trivial, but that they are now
solvable, particularly with emerging agent architectures and knowledge-graph approaches
that can impose structure and control on complex data environments.
One of the more provocative implications was the rise not just of agentic commerce, but
agentic customers in a world in which bots themselves become clients, complete with
identity verification and permissioning. It is a future that feels both imminent and faintly
unsettling, and one that raises as many governance questions as it answers.
Beyond AI, the broader programme leaned into themes of resilience, stablecoins, and
inevitably, agents, reflecting where industry attention is now consolidating: less on whether
the technology works, and more on whether financial institutions are structurally capable of
deploying it safely and at scale. And to add a little bit of magic, pun intended, the show
employed both a magician and a caricaturist to lighten the tone for when some of the topics
got heavy.
On the demo floor, FinovateEurope 2026 named three Best of Show winners. R34DY was
recognised for ABLEMENTS, a platform designed to accelerate AI-driven modernisation in
banks while reducing IT costs. Serene won for turning compliance into a growth lever, using
insights to optimise collections, reduce arrears, and safely expand lending. Tweezr
completed the trio with a developer-focused solution aimed at speeding up time-to-market
and boosting productivity across both legacy maintenance and modernisation.
Taken together, the winners reflected a broader truth about this year’s event: the most
compelling stories were not about shiny interfaces, but about making the slow, expensive,
and operationally painful parts of financial services work better. Showing that the core
opportunities of Fintech are still very much obvious, even in an AI-driven world.
