ING deploys AI mortgage agent to speed document checks

ING is deploying an AI mortgage agent in its live mortgage environment in the Netherlands after a pilot in March. The software will gather and check documents, move cases between internal systems and present likely outcomes to staff.

The agent scans applications to identify required paperwork, assembles and verifies documents and routes cases across ING systems. It flags probable results and next steps, while a human employee conducts the final credit assessment and approval.

Tom Degen, head of mortgages at ING Netherlands, commented: “With the agentic mortgage assistant, we are taking the next step in supporting mortgage applications to deliver faster decisions and clearer outcomes for customers and brokers.” He said the technology will allow staff to focus on more complex applications and direct contact with brokers.

The development forms part of a wider ING programme on agentic AI. Marnix van Stiphout, ING chief operating officer, previously described plans to use digital agents to collect client data and perform credit checks so customers would not always need to start an application by speaking to a person. Daniele Tonella, the bank’s chief technology officer, has identified five AI priority areas: customer identity checks, call centre support, wholesale customer due diligence, retail hyper-personalisation and internal engineering tools.

ING has said buying a home is one of the most important financial decisions people make and that much waiting for mortgage decisions happens while documents are gathered, checked and moved between systems and experts. The bank expects some operational roles to be redeployed as AI is implemented and has estimated that certain operational tasks could be done with about 25% fewer people; it plans to use capacity for growth and more complex work.

ING projects its AI investments will generate around €500m of additional value annually from 2030 onward. The live deployment will be monitored for effects on processing times, error rates and the balance between automated handling and human oversight as brokers and borrowers assess its impact.