Financial institutions are rapidly evolving with AI and the next frontier is about bringing AI agents to business users. According to speakers at DI’s BankTech event in Stockholm, this productivity leap is powered by a convergence of technologies and a shift from isolated innovation to systemic acceleration.

Who’s leading the way in banking AI adoption? According to Evident AI maturity global index, the top 10 names are all North American. European and Asian banks risk falling behind even though the Nordic region is in an enviable position when it comes to its high level of digitalization, strong platforms, robust regulation and highly educated workforce.

Among the tangible outcomes that AI leaders are delivering, you typically find advantages related to productivity gains. It’s about automating development workflows, accelerating customer processes, and utilizing AI to handle fraud. These are undeniably valuable advancements, but not yet transformative. To integrate AI into core operations, financial institutions must embrace deeper, structural change.

A three-vector approach to accelerating AI

How do you transition from experimentation to scaled value creation then? Scaling is very hard: Despite $30–40 billion in enterprise investment into generative AI, a 2025 MIT report uncovers a surprising result in that 95 percent of organizations are getting zero return.

One thing is clear: Firms that industrialize AI now will define the next decade’s platforms and utilities. Cognizant believes in a three-vector approach to accelerating AI, where successful businesses are on the brink of industrialization of AI: 

• Now: AI-led productivity to drive growth. Amplify human performance with AI-augmented workflows to create new customer experiences and pricing advantages through cost efficiency.
• 2026–2029: Industrializing AI to achieve operational resilience. Build scalable, secure and governed AI foundations to detect anomalies, ensure continuity and manage risk intelligently, preferably building around value streams.
• 2030 and beyond: Agentifying the enterprise to produce business agility. Empower systems and processes to act autonomously to enable faster, adaptive decisions.

The path to efficiency

Cognizant is currently applying multi-agent orchestration across the operating model in organizations – because when generative AI models generate predictions, AI agents actually do things. The fusion of these capabilities – creating multi-agent systems that both understand and act – is where the real transformation begins.

There are no shortcuts around data though, as usual; the fundamental layer supporting the agents comes down to data layers. Banking is, at its core, a data-driven industry. As AI becomes increasingly autonomous and less reliant on human oversight, the quality of data underpinning these agents becomes even more critical. Cognizant and AWS are thus working with clients to strengthen their foundational data layers.

Impacting operating models

With agents in place, you’ll have a hybrid work force, with humans and agents working in tandem. It’s important to give the agent approximately 5–10 tasks and apply the agents to certain functions and value streams of the bank, vertically and horizontally. A good example is preventative risk control where agents perform the heavy lifting, and the human workforce is empowered to take on more strategic tasks.

So how many agents will a major bank deploy? Estimates suggest between 5,000 and 10,000 agents – and they’re arriving fast. It’s important to understand that AI isn’t a tech level added to the existing stack; AI impacts operation models. Ultimately, AI has the potential to make or break a business. As one speaker at BankTech noted: “Non-compliance”