AI in Banking: Beyond the Hype—Are We Ready for Systemic Change?
Lasse Mäkelä (Multitude) & Petri Nikkilä (Openbank & Santander)
Santander manages Openbank, a cloud-based digital bank active in six countries, and a global consumer finance business collaborating with major partners like Apple and Amazon on checkout lending. AI applications focus on internal productivity, customer service via chatbots, and revenue generation, though the latter remains underdeveloped. Key challenges include data quality, regulatory demands for transparent AI models, and investor pressure prioritizing cost savings over innovative revenue streams. The rise of agentic commerce highlights new trust dynamics between customers, AI agents, and financial providers.
"The greatest potential of AI isn't just in improving productivity or customer service, but in building new revenue models and transforming entire ecosystems by becoming a trusted partner in agentic commerce—executing and financing complex transactions seamlessly for customers."
Summary
- Santander focuses AI in three areas: internal productivity, customer service, and new AI-driven revenue models, with the last being most transformative but least developed. - AI chatbots often lead to poor customer experiences, and true AI potential lies in agentic commerce that automates complex tasks like holiday planning and payments. - Data organization and strict regulatory requirements are major obstacles to scaling AI in banking, alongside pressure to show short-term cost savings. - A notable AI success at Santander is a broker platform that uses AI for stock recommendations, which outperforms human analysts about 60% of the time. - Building trust with regulators and customers is crucial, especially for automated AI models and AI agents managing payments or financing, requiring transparency and proof of fairness.
Article
AI in banking: beyond the hype – are we ready for systemic change?
Banking sector struggles to balance short-term AI cost savings with transformative revenue growth
Lasse Mäkelä from Multitude and Petri Nikkilä from Openbank and Santander delivered a thought-provoking examination of artificial intelligence in banking at the Nordic Fintech Summit, challenging industry assumptions and highlighting missed opportunities for transformative change.
Speaking at the event on Wednesday, 14th of May 2025, Nikkilä, who oversees Santander's cloud-based digital bank Openbank and its global consumer finance operations, presented a candid assessment of AI's current implementation in banking.
"AI has three value buckets in banking: internal productivity, customer interaction, and revenue-building with new business models. The third category has the biggest change potential but gets the least focus," Nikkilä explained during the session.
The false promise of AI chatbots
Nikkilä was particularly critical of the banking sector's fixation on AI chatbots, which he characterised as merely making poor customer experiences "slightly less unpleasant."
"Chatbots are an unpleasant experience for customers; the most common request is to connect to a human advisor. If this is the pinnacle of AI in customer service, then our industry has failed," he said.
This focus on incremental improvements rather than transformative applications stems partly from pressure from investors and analysts who demand immediate cost savings that can be quantified in quarterly reports.
The untapped potential of agentic commerce
The most compelling vision Nikkilä presented was around "agentic commerce" – AI agents that could autonomously handle complex tasks like organising family holidays and managing associated payments and financing.
"Agentic commerce powered by AI could revolutionise organising family holidays, removing search engines and redefining the ecosystem where trusted partners, not intermediaries, own the purchasing experience," Nikkilä said.
In this scenario, consumers could instruct an AI agent to plan within specific budget parameters, including financing options, fundamentally altering the consumer journey and potentially disrupting major players like Google.
Success stories amid challenges
Despite the challenges, Nikkilä highlighted a successful AI implementation at Santander – an AI-powered stock recommendation system on their broker platform that outperforms human analysts.
"With AI, by collecting all of the global data for any company you want to analyse, plus the analyst recommendations, we are right about 60% of the time," he explained, expressing surprise that more banks haven't implemented similar solutions.
Building trust in AI systems
The dual challenge of establishing regulatory trust and consumer trust emerged as critical themes. Banking regulators demand transparent, explainable models rather than "black box" AI systems.
"ECB, for example, doesn't approve any models that are black box models that they don't understand deep down," Nikkilä noted, explaining that regulators want to be able to examine models in detail before approval.
Additionally, as AI agents gain authority to make financial decisions on consumers' behalf, financial institutions must establish themselves as trusted partners in this new ecosystem.
Obstacles to AI transformation
Nikkilä identified two primary obstacles to meaningful AI implementation in banking: poor data organisation and short-term financial pressures.
"All of us big banks have data and all of us have the same challenges: it's not organised in a very good fashion," he said, while also noting that regulatory requirements demand near-perfect accuracy.
As the Nordic Fintech Summit continues through Thursday, 15th of May 2025, the candid assessment offered by Nikkilä provides a roadmap for financial institutions seeking to move beyond superficial AI implementations toward truly transformative applications that could reshape the industry's future.
Part of Nordic Fintech Summit