San Francisco-based Addy AI has launched what it says is the first fully agentic ChatGPT app built specifically for mortgage pre-underwriting, allowing lenders and LOs to move parts of the pre-qualification and file review process directly into ChatGPT.
The app, which went live Wednesday, integrates Addy AI’s mortgage-focused AI agents into ChatGPT workflows. The platform can review borrower documents, analyze loan scenarios, identify missing conditions, and generate structured pre-underwriting findings in roughly five minutes.
The larger strategic shift may be happening even earlier in the mortgage process itself.
Mortgage technology firms are increasingly betting that borrowers will begin their financing journey inside AI assistants before ever visiting a lender website, completing a lead form, or speaking directly with a loan officer.
“Borrowers are already starting their home-buying journey inside AI,” said Michael Vandi, CEO of Addy AI. “The opportunity for lenders is to meet that intent the moment it happens. With Addy AI in ChatGPT, a borrower’s first question can become the beginning of a real mortgage workflow, and our AI Agents can help pre-underwrite the loan in 5 minutes.”
For mortgage professionals, the bigger story may be less about replacing underwriting staff and more about compressing the time from borrower inquiry to an actionable loan review.
The company says the integration allows LOs and mortgage teams to use ChatGPT as a front-end conversational workflow tied directly into pre-underwriting preparation. Borrowers can ask questions about affordability, qualification requirements, required documentation, and next steps, while Addy AI structures the information and prepares the file for lender review.
The Addy AI system supports workflows including:
- Mortgage pre-underwriting
- Borrower pre-qualification
- Income, asset, and credit review
- Missing document identification
- AI-generated loan processing checklists
- Loan scenario analysis
- Custom lender AI agents inside ChatGPT workflows
AI conversations inside mortgage lending are shifting from broad experimentation toward operational workflows tied directly to speed, borrower conversion, and production efficiency.
Last month, Rocket Close implemented AWS-powered AI tools designed to reduce mortgage document review times from roughly 10 hours to under two minutes, highlighting how aggressively lenders and mortgage technology firms are pursuing workflow compression and faster loan execution.
At the same time, the industry is still navigating what meaningful AI adoption actually looks like in practice. A recent AD Mortgage survey found that while brokers are rapidly adopting AI tools across document processing and borrower communication workflows, many remain undecided about which technologies will ultimately deliver measurable production value.
Addy AI positions its platform specifically around the earliest stages of the mortgage process, where lenders often lose borrowers before a formal application is ever completed.
For LOs, the longer-term implication may be that “speed-to-lead” is quietly evolving into “speed-to-intent,” with borrowers increasingly forming financing opinions and exploring qualification scenarios inside AI platforms before formally entering a lender’s pipeline.
Addy stated that speed matters to the borrower, and that many consumers now begin their financing research via AI assistants, search engines, and digital channels long before entering a lender’s traditional workflow.
The company emphasized that the platform does not replace licensed mortgage professionals or formal underwriting decisions. Instead, Addy AI says its AI agents are designed to assist with document review, file preparation, and issue identification prior to underwriting.
Addy AI also said it worked with support from the OpenAI team during the launch process.
“ChatGPT apps are becoming a new distribution layer for software,” Vandi said. “For mortgage, this means the borrower’s first AI conversation can become the beginning of a real, structured loan workflow.”
That “distribution layer” concept may ultimately become one of the more important implications for mortgage technology providers. If borrowers increasingly begin their financing journey inside conversational AI platforms, mortgage software itself could gradually shift away from traditional standalone dashboards and toward embedded conversational workflows where borrowers, loan officers, and AI agents interact simultaneously.
According to Vandi, more than 50 loan officers have already signed up to use the platform since launch.
*This article was primarily written by a human author. AI tools were used in a limited capacity for research assistance or light editing.
