%201.webp)
Blend has released early production results for its Autopilot lending system, based on more than 50,000 live production loans on its Home Lending platform.
Lenders using Autopilot's pre-underwriting agent recorded pull-through rates 10% to 15% higher than before adoption. Loan-cycle times were cut by two to four days, while an average of 4.5 hours of fulfilment tasks were automated for each loan.
Blend estimated the change reduced fulfilment costs by USD $600 per funded loan. The comparison used 24 lender and loan-type cohorts measured against their own pre-Autopilot performance, covering more than 175,000 loans.
The results come from live borrower files rather than pilot programmes or simulations. That distinction matters in mortgage lending, where vendors have promoted artificial intelligence tools for tasks such as document review and workflow support, but production evidence has often been limited.
Live deployment
Autopilot became commercially available to lenders on Blend's Home Lending platform on 1 July. Six lenders have already signed contracts that include the system.
Blend describes Autopilot as an agent-based system that conducts pre-underwriting work in real time. It reviews mortgage applications as they arrive, identifies missing or inconsistent information, and follows up with borrowers while they are still engaged in the process.
That focus on the opening stage of an application reflects borrower behaviour on the platform. More than half of borrowers apply for a mortgage outside business hours, and more than 90% of those who submit an application do so within 24 hours of starting.
For lenders, that first day can be decisive. Delays at that stage can leave applications incomplete and increase the risk that borrowers move to another provider before a loan team responds.
Blend argues that most existing AI tools in lending still speed up individual steps in a process that remains sequential. Autopilot, by contrast, works files in parallel and in real time, changing both the cost and pace of loan origination.
Nima Ghamsari, Co-Founder and Head of Blend, set out the company's view of how AI should be applied in lending.
"I think the biggest mistake people make with AI is trying to be too incremental. Our approach is to ask: What would this industry look like if you designed it from the ground up around agents?" Ghamsari said.
He added that the company is now seeing that approach in live loan production.
"That's no longer a thought experiment. Autopilot has run on more than 50,000 live production loans, and the economics are already changing," Ghamsari said.
First 24 hours
The pre-underwriting agent operates within the first 24 hours after a borrower begins an application. During that period, the system is intended to complete early file work before a staff member reviews the loan, reducing later rework and borrower callbacks.
That marks a notable shift from traditional mortgage operations, where staff often make the first pass through an application and identify issues only after the file has moved to a later stage. Blend said cleaner files reaching processing and underwriting can reduce the number of late-stage conditions and shorten the path to a decision.
Ghamsari said the system changes the order in which work is done.
"Instead of a person taking the first pass and an agent checking the work, Autopilot agents are working behind the scenes before a human ever touches the file. Rules-based automation could never do that, because rules only work on predictable files, and underwriters spend all their time on the ones that aren't," he said.
He said the aim is to free loan teams to focus on direct borrower contact rather than manual file preparation.
"By the time it reaches a loan team, the work is done, which gives them their time back to spend with borrowers instead of paperwork," Ghamsari said.