Danske Bank is piloting a Model Context Protocol (MCP) Server that offers a practical example of agentic AI in banking. The solution lets corporate customers connect AI agents – like Claude or ChatGPT – directly to their own banking data through the bank’s Premium APIs, using their own credentials. According to Danske Bank, this is the first pilot of its kind from a Nordic bank, enabling customers to access and view account information through AI tools under their own control.
A New Interface for Business Banking
The pilot lets corporate customers type natural-language requests – asking for their account balance or requesting a three-month cash-flow forecast, for example – with responses generated from the customer’s own banking data via the bank’s Premium APIs, as FinTech Magazine reports. This positions MCP as a potential third channel for business banking, sitting alongside conventional digital banking and traditional APIs.
MCP itself is an open standard that gives AI applications a uniform way to plug into external tools and enterprise systems. It sits above existing APIs, so that AI agents can discover and use approved services without a bespoke integration for each platform. For corporate finance teams, that could ease the friction of switching between bank portals, treasury systems, accounting software and ERP platforms.
Three interfaces for business banking: online banking, traditional APIs and AI agents via MCP
| Online banking | Traditional APIs | AI agent via MCP | |
| Who uses it | The customer, by hand | The customer’s IT systems | The customer’s AI agent |
| How requests are made | Clicks and forms | Code written for each system | Natural-language prompts |
| Integration effort | None | One build per system | One standard for any compatible AI tool |
Christie H. Kristensen, Danske Bank’s Strategic Integrations Partnership Director, says the bank sees a growing shift toward MCP across accounting and treasury platforms, and describes it as a strong signal of where the financial ecosystem is heading. The objective of working with an ecosystem of service providers is to help customers unify financial data that has historically been scattered across disconnected systems – positioning Danske as the first Nordic bank giving customers this kind of freedom to orchestrate their financial operations through the agentic AI tools of their choice.
Part of a Broader Strategy
The pilot sits within Danske Bank’s “Forward ’28” strategy, which prioritizes digital services, advisory, sustainability, and simple, efficient and secure banking, with plans to scale AI-enabled solutions across customer journeys. Sandra Schaldemose Sommerlund, the bank’s Tribe Lead of Customer & Sector Integration, said the goal is to make Danske’s services available whenever and wherever customers need them, seamlessly integrated into whichever tools they choose.
Some figures behind that ambition:
- Danske has moved more than 30% of its applications to the public cloud since 2023.
- The bank aims for AI to handle more than 50% of customer inquiries by 2028.
- Planned annual investment in core technology, AI-enabled platforms and advisory has been raised from roughly DKK 4 billion to approximately DKK 4.5 billion (about US$694.5 million).
- Targets for 2028 include return on equity above 14.5%, a cost-to-income ratio no greater than 43%, and total income of roughly DKK 63 billion (about US$9.7 billion).
What does agentic AI in banking mean for other banks?
What does agentic AI in banking mean for other banks?
Danske’s pilot is about corporate treasury, but the pattern – a customer’s own AI agent calling the bank’s services – is unlikely to stay there. Gartner predicts that by 2028 at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from zero in 2024.
Follow the logic one step further. Once an agent can build a cash-flow forecast, the next request is predictable: “Book a meeting with our relationship manager next week to discuss a credit line.” A bank that can’t accept that request through an API sends the customer back to a web form or a phone queue.
This doesn’t make the branch obsolete. Advice and complex decisions still happen between people. What changes is the logistics around them: finding the right advisor, choosing a slot, sharing context before the meeting. Banks that already run these steps on a connected appointment booking platform – one that integrates with their mobile app, CRM and chatbots – will find it much easier to open them up to AI agents when customers start asking.
Is your branch network ready for AI agents? Download our free e-book, Redesigning TRUST – The Role of Bank Branches in the Digital Age.