Sela, a startup building voice AI agents for the mortgage industry, has secured $21 million in funding across its Seed and Series A rounds. The investment was led by Costanoa Ventures, with participation from Emergence Capital. This capital injection arrives as the company reports significant traction in a sector where AI is often discussed in abstract terms but rarely applied to high-stakes financial operations.
The company’s core product is a voice AI agent designed to handle the heavy lifting of mortgage origination. Rather than acting as a simple chatbot, these agents are built to manage complex borrower interactions, build rapport, and guide individuals through the loan process. They only escalate to a human loan officer when necessary. According to the company, this technology is currently facilitating over $1 billion in mortgage originations every month.
The financial growth metrics are equally aggressive. Sela reached a $10 million annualized run-rate in just 18 months. Today, six of the ten largest independent mortgage banks in the United States have integrated the technology into their production workflows. This level of adoption among industry leaders suggests that lenders are prioritizing tools that can directly impact their bottom line in a challenging interest-rate environment.
David Cheng, a partner at Costanoa Ventures, notes that the current market pressure is driving this demand. As he puts it, “Every lender’s P&L comes down to the same two numbers: conversion and cost per funded loan, and both are under more pressure right now than at any point in a decade.” Sela is positioning itself as a solution to these specific constraints rather than a general-purpose AI tool.
CEO Nate Becker emphasizes that the company’s internal success metrics are tied to tangible outcomes rather than engagement time. “We measure success in borrowers reached, productive conversations held, and loans closed, not minutes used,” Becker says. This focus on unit economics is a central part of their pitch to institutional lenders.
To validate this approach, Sela points to internal A/B testing data. In one trial involving more than 10,000 borrowers, the company reported a 9% increase in lead-to-lock rates, which they claim resulted in 40% higher profit for the lender. In a separate test against a competing voice AI solution across 7,000 leads, Sela claims to have outperformed the alternative by 41%.
It is important to note that these performance metrics are company-sourced. As is standard practice in the industry, these self-reported figures should be viewed with caution until they are validated by long-term, third-party audits. While the reported growth is significant, the company’s ability to maintain this performance as it scales from its current team of 17 to a planned 50 employees remains the primary test of its model.
Sela’s trajectory mirrors a broader trend in the AI sector toward vertical agent economics. This shift is also visible in the recent $240 million raise for Owner, a company that similarly focuses on generating measurable revenue within a specific, high-value vertical. Investors are increasingly moving away from general-purpose AI demos and toward companies that can prove repeatable revenue in complex, niche industries.
By focusing exclusively on the high-friction world of mortgage origination, Sela is attempting to turn AI into a reliable utility. The company is not trying to solve every problem for every business; instead, it is designing and refining agents for a specific set of tasks. If they can sustain their current conversion metrics, they may provide a blueprint for how AI can successfully automate the difficult, repetitive work that has historically required large human teams to manage.
Ultimately, the success of this model will depend on whether these agents can continue to deliver consistent results as they are deployed across a wider range of mortgage portfolios. For now, the company is betting that the future of enterprise AI lies in deep, vertical integration rather than broad, horizontal capability.