In the first post of our AI in Payments series, we asked who might be best positioned to capture the AI opportunity in payments. The answer was not simply banks versus fintechs, processors versus Big Tech, or incumbents versus challengers— it was whether an organization can adopt AI in a way that is fast, secure and genuinely transformative.
That leads to the next question.
Where should AI deliver value first?
Should financial institutions use AI primarily to reduce costs and improve operations? Or should they focus on creating new products, stronger customer experiences and new revenue streams?
We put this question to Tomas Janitor, Head of Business and Innovations at Raiffeisen Bank’s RPC; Alessandro Hatami, Managing Partner at Pacemakers.io; and Emanuel Doneda, Partner at Deloitte.
The discussion suggested the answer is both — but in sequence, not simultaneously.
Watch the video:
Efficiency comes first
For most institutions, the immediate AI opportunity is operational.
Back-office processes, manual workflows and internal productivity are areas where AI can already deliver measurable impact. Reducing effort, accelerating routine tasks and lowering operational costs are often the most realistic short-term goals.
But this first phase is about more than optimization. It is also how organizations build practical experience with the technology.
Internal deployment gives institutions a clearer understanding of where AI delivers value — and where its limitations remain. That experience becomes critical before moving into more ambitious customer-facing use cases.
AI is not a universal solution
Operations, compliance, servicing, risk and product development all work differently. Each process has its own constraints, objectives and data realities. Successful AI adoption therefore requires institutions to rethink processes individually and ask a more focused question:
What specific problem are we solving?
This sounds obvious, but many organizations are still at an early stage. In one survey referenced during the discussion, only a minority of employees had regular access to AI tools, and even fewer used them daily. Before AI transforms customer experience, many firms still need to integrate it meaningfully into their own operations.
Experimentation matters — but so do safeguards
Another recurring theme was realism.
Many AI use cases in financial services are still being tested at scale, which makes experimentation essential. Institutions need room to test boundaries, learn quickly and adapt their approach as the technology evolves.
At the same time, they need safeguards. Failed AI projects should not create operational, regulatory or reputational risk for the wider organization.
This balance between experimentation and control may become one of the defining management challenges of the AI era.
From optimization to growth
Operational improvement may be the natural starting point, but it is unlikely to be the end goal.
The larger opportunity lies in using AI to improve customer experience, create more intelligent services and develop new forms of value. But institutions will only get there if they first build the governance, operational understanding and internal capability required to scale AI responsibly.
In that sense, the debate is not really about choosing between efficiency and growth.
Efficiency is the first phase of AI maturity.
Growth is what follows.
This is the second post in our AI in payments series from Il Salone. Over the next posts, we will explore the questions financial institutions are now facing most urgently: why so many projects fail to deliver impact, how success should be measured, how the industry may evolve over the next few years, what projects are delivering real lessons today, and what leaders see as the next step in their own AI journey.
-2.jpg)
