Weekly Real Estate News
Summary
Blend reports ten lenders have signed on to Autopilot and the mortgage AI system handles more than 3,500 new loans weekly; claimed time savings have not been independently verified.
Blend says ten lenders have signed on to its Autopilot mortgage automation system, with the software now working on more than 3,500 new loans a week. The figures, disclosed Monday, offer a measure of adoption as mortgage lenders experiment with artificial intelligence.
Whatcom Educational Credit Union in Washington and Gate City Bank are among the recently named institutions. Blend did not identify every lender in the group or say how many have moved from contract signing to full production. Its 3,500-loan figure measures files touched by the software, not mortgages underwritten or closed without human intervention.
The announcement comes six months after the system began handling live loans and three months after commercial availability, according to Blend. At the current pace, 3,500 new files weekly would equal roughly 182,000 over 52 weeks, provided volumes remained unchanged. That is an annualized activity measure, not a completed-loan forecast.
Autopilot uses software agents to examine borrower documents, review income, follow up on missing items and assist with pre-underwriting and closing quality control. It operates within Blend’s home-lending platform. Lenders remain responsible for credit decisions, compliance and oversight of vendors and automated processes.
Blend cited earlier production results showing loan pull-through 10% to 15% higher, cycle times two to four days shorter and an average of 4.5 hours of fulfillment work automated per loan. The company did not publish the underlying sample or independent audit with Monday’s release. Those figures are company-reported results, not savings every lender should expect.
Onity Mortgage, an early adopter, said its pre-underwriting workflow can review documents when borrowers upload them outside business hours and request replacements if submissions do not meet requirements. That could reduce time lost waiting for staff review, although it does not establish that automated review can resolve every exception.
For loan officers, the practical question is whether the system catches incomplete files early enough to avoid costly last-minute conditions. For compliance teams, it is how the lender records an automated action, checks for errors and preserves human review when borrowers challenge results.
Blend is competing in a crowded field. FirstClose and Cotality recently announced a separate home-equity processing integration. The products are different, and neither company’s performance claims establish an industry-wide benchmark.
More useful disclosures would include the number of loans completed using Autopilot, implementation costs, exception rates and verified time savings across lenders. Those measures will determine whether the growing workload represents a lasting change in mortgage production.
