Better Automates 1.89M Mortgage Calls With ElevenLabs Voice Agents.

Online mortgage lender Better has deployed ElevenLabs' conversational AI platform to handle nearly 2 million customer calls, automating routine borrower interactions from eligibility checks to rate locks and freeing loan officers to focus on complex cases.

The company built Betsy, an AI loan assistant powered by ElevenAgents, to manage phone conversations that traditionally required human agents. According to Better, the system handled over 1.89 million calls in 2025, with 35 percent never requiring escalation to a human consultant.

Better reports the AI assistant doubled its lead-to-lock conversion rate while saving 1,600 hours of loan officer time each month. The system handles qualification calls, pricing inquiries, and outbound follow-ups, allowing licensed consultants to concentrate on cases requiring nuanced judgment or relationship building.

The mortgage industry has long struggled with high-volume, low-complexity calls that tie up licensed professionals. By routing routine inquiries to a conversational agent, Better aims to compress decision timelines and improve rate competitiveness without adding headcount.

Betsy operates over the phone, a channel where many mortgage customers still prefer to transact. The agent fields questions about eligibility, explains pricing scenarios, and walks borrowers through rate lock processes using natural language.

The Better deployment adds to a growing list of financial services firms adopting ElevenLabs' voice AI. Revolut selected ElevenAgents to bolster customer support in January, while Klarna reported a 10x reduction in time to resolution after implementing the platform, as previously covered.

ElevenLabs recently closed a $500 million Series D at an $11 billion valuation, reflecting investor confidence in voice AI's enterprise potential. The company has also announced partnerships with Deutsche Telekom and Deloitte to deploy agents at scale across telecommunications and consulting.

Better's use case is notable for the volume and sensitivity of interactions. Mortgage qualification involves regulated disclosures, complex eligibility logic, and financial decisions that can determine whether a borrower secures a home. The company's claim that more than a third of calls never escalate suggests the agent handles a significant share of standard scenarios without human intervention.

If Better's results hold across a broader sample, the implications for mortgage operations are significant. The industry processes millions of loans annually, each requiring multiple customer touchpoints. Automating even a fraction of those calls could reshape cost structures and speed up closings.

The 2x conversion improvement is particularly striking. Faster response times and 24/7 availability could reduce drop-off rates, especially for borrowers shopping multiple lenders. Whether that metric reflects the agent's performance or broader operational changes at Better is unclear from the announcement.

Better did not specify which portions of the mortgage workflow Betsy handles, how escalations are triggered, or whether the agent operates independently or as part of a hybrid model. The company also did not disclose error rates, customer satisfaction scores, or regulatory review processes.

As voice AI moves deeper into regulated industries, questions around oversight, transparency, and fallback mechanisms will likely intensify. For now, Better's deployment offers a high-profile example of conversational agents handling real-world financial transactions at scale.