Better.com Launches Betsy With ElevenLabs Technology, First AI Voice Loan Assistant For Mortgages

Better partners with ElevenLabs to make AI assisted loans accessible through the first voice-based loan assistant built exclusively for the mortgage industry

ElevenLabs, the AI audio research and deployment company, released a new enterprise case study highlighting how Better.com, a leading AI-native home finance company, uses ElevenLabs Agents to automate mortgage conversations at scale. This empowers Better.com to provide significant cost savings, higher conversion, and a faster borrower experience while meeting the demands in a highly regulated industry.

Better.com uses ElevenLabs Agents as the voice interface for Betsy™, its generative AI loan assistant, to handle nearly 100,000 mortgage-related phone calls per month, automating 35.5% of borrower inquiries end to end. In 2025 alone, Betsy placed 1.89 million inbound and outbound calls, saving Better's loan officers more than 1,666 hours of human time each month.

"We built Better to be AI-native from day one. By combining Tinman® with ElevenLabs' technology, we've scaled the first voice-base AI loan assistant built exclusively for the mortgage industry," said Vishal Garg, Founder and CEO of Better.com. "AI-voice automation through Betsy™ has allowed us to move routine customer interactions out of manual workflows, helping us reduce the cost to originate while offering borrowers immediate and 24/7 support."

These gains have allowed Better's licensed consultants to shift away from repetitive follow-up tasks and focus on higher-value advisory work - helping borrowers navigate complex scenarios and make informed financial decisions.

Measurable Impact on Mortgage Economics

According to the case study, Better's use of AI voice automation has delivered material business results across its consumer mortgage business, including:

• 41% reduction in average cost to originate
• 100% increase in lead-to-lock conversion in 2025
• More competitive borrower rates, driven by operational efficiencies passed directly into pricing

By moving routine eligibility checks, pricing conversations, and rate locks to an AI voice agent, Better has improved both borrower experience and unit economics without compromising compliance.

Better initially launched its voice experience using a speech-to-speech model but transitioned to a modular pipeline combining speech-to-text, a large language model, and text-to-speech with ElevenLabs Agents to achieve lower latency, greater control, and higher reliability. A key factor in selecting ElevenLabs was platform stability under high tool-call volume. In production, a single borrower conversation may trigger dozens of internal system calls. For financial services workflows that span multiple systems and regulatory checkpoints, this level of reliability at this scale is critical.

Betsy™ is layered on top of Better's proprietary loan engine, Tinman®, which computes pricing, eligibility, and approvals. ElevenLabs Agents serve as the controlled voice interface, communicating Tinman's outputs to borrowers. All regulated steps - including credit pulls, authentication, rate locks, and transfers to human consultants - require explicit consent and adhere to state licensing requirements.

Orchestration runs entirely within Better's environment through an MCP-based multi-agent system connected to Tinman tools, while calls terminate at a single ElevenLabs Agent endpoint. This architecture ensures determinations remain within Better's systems, with ElevenLabs providing a consistent, natural voice experience.