Blend Autopilot
AI agent · Workflow / back-office agent · Operations & supply chain by Blend
Put pre-underwriting on Autopilot to close loans faster and reduce costs.
Blend Autopilot is an AI-native pre-underwriting agent that automates document review, income calculation, and guideline checking for mortgage and home equity lenders. It helps financial institutions close loans up to 4 days faster and save $793 per funded loan by processing files 24/7.
Agent specs
The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.
Category: Lending
- Integrations
- Blend platform, Fannie Mae, Freddie Mac
- Data sources
- Applications, documents, borrower activity, and loan data
- Deployment
- Integrated within the Blend platform
- Time saved per file
- 4.5 hours
- Cost savings per funded loan
- $793
- Reduction in time to close
- 2-4 days
- 4.5 hours saved per loan file
- 2-4 days faster time to close
- $793 saved per funded loan
- 10-15% increase in funded loans
- 24/7 automated document review and income calculation
Overview
Blend Autopilot is an AI-native pre-underwriting agent designed for financial institutions to automate mortgage and home equity loan origination. It works continuously to review documents, calculate qualifying income, and check files against investor guidelines as soon as information is submitted. By surfacing missing requirements while borrowers are still engaged, it helps lenders close loans two to four days faster and save an average of 4.5 hours per file. The agent is built with explainable AI, logging every action and citing specific guidelines for its findings to ensure transparency for human underwriters. It integrates directly into the Blend platform, utilizing existing lending context and guardrails to keep loans moving forward 24/7.
Pre-underwriting Automation
- Document Review: The agent reads, classifies, validates, and extracts key data from borrower documents as soon as they are submitted.
- Income Calculation: It applies specific rules to calculate qualifying income and provides the mathematical breakdown for every calculation.
- Guideline Checking: The system validates loan files against Fannie Mae and Freddie Mac guidelines, as well as lender-specific overlays and custom rules.
- Automated Follow-ups: It generates targeted follow-up requests for missing information and automatically re-reviews new data when it arrives.
Workflow Efficiency
- 24/7 Processing: The agent works on files as information comes in, including after business hours when approximately half of borrower activity occurs.
- File Maintenance: It updates the loan file, clears resolved items, and surfaces items that still require attention to keep the process moving.
- Explainable Findings: Every finding includes a citation back to the applicable investor or lender guideline for transparency.
- Auditable Actions: All actions taken by the agent are logged to provide a clear audit trail for human reviewers.
Human Control and Oversight
- Configurable Capabilities: Lenders choose which specific capabilities to enable and which loan officers have access to the agent.
- Human-in-the-Loop Decisioning: The agent assists with pre-underwriting work but does not make final credit decisions, which remain with the human team.
- Action Controls: Users can determine whether certain actions happen automatically or are held for manual review.
What it does
- Document review\nIncome calculation\nGuideline checks\nBorrower follow-ups\nApplication data updates
Published results
- 10-15% more funded loans\n2-4 days faster to close\n4.5 hours saved per file\n$793 saved per funded loan
Use cases
- Lenders use Autopilot to automatically review and extract data from borrower documents as they arrive to speed up the initial file review.
- Underwriters use the agent's automated income calculations and math breakdowns to verify qualifying income more efficiently.
- Financial institutions deploy the agent to validate loan files against Fannie Mae and Freddie Mac guidelines 24/7, even after business hours.
- Loan officers use automated follow-up generation to request missing information from borrowers while they are still actively engaged in the process.
Specifications
- Integrations
- Blend platform, Fannie Mae, Freddie Mac
- Data sources
- Applications, documents, borrower activity, and loan data
- Human control
- Assists with pre-underwriting; credit decisions remain with the human team; actions can be held for review
- Deployment
- Integrated within the Blend platform
- Time saved per file
- 4.5 hours
- Cost savings per funded loan
- $793
- Reduction in time to close
- 2-4 days
- Increase in funded loans
- 10-15%
- Supported loan types
- Mortgage, Home Equity
- Operating hours
- 24/7
- Guideline support
- Fannie Mae, Freddie Mac, Lender Overlays
How it works
- Document Ingestion and Review: The agent reads, classifies, and extracts data from borrower documents immediately upon arrival.
- Income Calculation: It applies rules to calculate qualifying income, showing the math for every result.
- Guideline Validation: The file is validated against investor guidelines and lender-specific overlays.
- Resolution and File Update: The agent generates follow-ups for missing data and updates the loan file automatically.
FAQ
When does Autopilot start working on a loan?
It responds to loan activity, including application completion, document uploads, borrower responses, and AUS submissions, often starting before the file reaches fulfillment.
Which lending guidelines does Autopilot support?
It supports Fannie Mae and Freddie Mac guidelines, as well as lender-specific overlays and custom rules.
How much control do I have over Autopilot?
Lenders can choose which capabilities to enable, which loan officers use it, and whether actions are automatic or held for review.
Added by AntSpace from public sources (blend.com).