Blend Launches Autopilot Analytics Agent

New AI agent gives lenders real-time visibility into borrower activity and funnel performance from application through close

SAN FRANCISCO--(BUSINESS WIRE)--Today, Blend (NYSE: BLND) announced the general availability of the Autopilot analytics agent, an AI agent that gives lenders instant visibility into borrower activity and funnel performance from application start through close.

By asking questions in plain English, lenders can identify bottlenecks, monitor conversion, and act on issues while loans are still moving through the pipeline.

Run this month’s pipeline, not last month’s report

Every borrower interaction generates signals about where loans are moving forward, slowing down, or being lost. Yet for most lenders, those signals remain buried in reports, analyst queues, and spreadsheet exports that arrive weeks after decisions could have been made.

The Autopilot analytics agent changes that. Lenders can ask questions about their Blend POS data in plain English and instantly receive dashboards, benchmarks, and alerts that surface issues while there is still time to improve outcomes.

Blend's network data shows that only 60-70% of retail mortgage application starts ever reach submission, yet many lenders have little visibility into where borrowers abandon the process. Those borrowers never appear in traditional pipeline reports, even though the lender already paid to acquire them.

The analytics agent gives lenders:

• Real-time pipeline visibility. Track borrower activity, stage-to-stage conversion, stalls, and drop-off from application through close.
• Proactive monitoring. Receive threshold alerts through email, Slack, or Microsoft Teams the moment key metrics change.
• Internal and external benchmarking. Compare loan officers and branches internally while benchmarking against anonymized peer segments across Blend’s network.
• Collaborative dashboards. Build, save, schedule, and share customizable dashboards using pre-built lending templates.

Ask your POS data directly

The analytics agent connects directly to a lender’s Blend POS data, turning natural-language questions into dashboards, charts, and insights in seconds. Teams can instantly understand how quickly borrowers complete applications, which lead sources generate the strongest results, where follow-up operations slow down, and how borrowers rate their experience across regions, branches, loan officers, and products.

Built on 50M+ applications, captured at the start of the funnel

Blend processes approximately $1.3 trillion in loan applications annually and has powered more than 50 million consumer loan applications over nearly 15 years.

Because Blend sits at the point where borrowers apply, it can measure behavior before submission: where applications stall, the cost of that friction, and what high-performing funnels look like. Those insights power benchmarking for individual lenders while providing context from across Blend’s customer network.

"Seeing your business should be as immediate as running it. The analytics agent gives lenders the visibility of a network that powers one in six U.S. mortgages, in time to act on it," said Nima Ghamsari, Co-founder and Head of Blend.

Available today

The Autopilot analytics agent is available today for customers using Blend Home Lending, including Mortgage and Home Equity. For more information, visit Blend’s Autopilot analytics page.

About Blend

Blend (NYSE: BLND) is the AI infrastructure for relationship banking. We help banks and lenders spend less time on process and more time with people. We do this by connecting the systems, guardrails, and workflows that powered $1.3 trillion in loans in 2025. Autopilot, our flagship AI product, runs on that infrastructure to read, analyze, and safely move loans forward in real time. The result: lenders save up to 4.5 hours per loan, cut 2 to 4 days from cycle times, and improve pull-through by a median of nearly 19%. By giving teams more capacity, Blend makes relationship banking possible at scale. Learn more at blend.com.