Yorkshire Building Society taps Covecta AI for underwriting

TL;DR:

  • Yorkshire Building Society has partnered with Covecta to bring agentic AI into its underwriting process.
  • The tools automate routine tasks and surface relevant information, keeping human underwriters in control of decisions.
  • It is a concrete example of a named UK mutual embedding AI in lending operations.

Britain’s mutual lenders are quietly putting AI to work in the back office. Yorkshire Building Society has partnered with Covecta to deploy agentic AI in underwriting — automating routine document review so underwriters can spend more time on complex cases and customer conversations.

Augment, not replace — the industry’s chosen framing

Both sides stressed that the technology is designed to support, not supplant, human judgement. Simone Fox, the society’s director of customer support, said the partnership frees underwriters to focus on “fair, informed lending decisions” while cutting the manual effort of reviewing paperwork. Covecta founder Scott Wilson put it more bluntly: financial institutions “do not need technology that replaces expertise; they need technology that amplifies it.” Covecta builds AI agents for banks, building societies and lenders across acquisition, screening, underwriting and monitoring, and says its platform is shaped around the sector’s regulatory and operational demands.

That “augment, not replace” message has become near-universal in financial-services AI announcements — reassuring for staff and regulators alike, though the harder questions about accountability and oversight tend to sit beneath the headline. The deployment fits a measured UK pattern, sitting alongside moves such as NatWest training all 60,000 staff in AI ethics and the broader push by lenders to modernise operations without spooking customers. It also lands the same day the FCA set out how it is rethinking AI regulation for exactly these kinds of automated decisions.

Looking forward

For UK borrowers, AI in underwriting promises faster decisions — but the value, and the risk, lies in how much weight lenders place on the machine’s output. The test for building societies will be holding the line on human oversight and explainability as the tools take on more of the workflow, particularly where automated decisions affect who gets a mortgage and on what terms.

Share this article