Better.com AI: How the Mortgage Disruptor Rebuilt Loan Origination Around Conversations

TL;DR#

Better.com (NASDAQ: BETR) is the publicly-traded digital mortgage lender that, after a turbulent SPAC debut, mass layoffs, and a CEO-on-Zoom scandal in 2021–2022, rebuilt its product around conversational AI — most visibly with Betsy, the company's AI mortgage assistant launched in 2024 and expanded in 2025. The company funded roughly $3.6B in loans in Q3 2024 and reported that AI-driven workflows now touch the majority of its origination funnel, including pre-approval, follow-ups, and customer service. Better's path-to-pre-approval — historically a 20+ minute online form plus follow-up calls — has been recompiled into a single conversational session that can return a pre-approval letter in under three minutes for eligible borrowers. The broader lesson for fintech intake-heavy companies is not "build your own Betsy" — it is that automated lead qualification software in mortgage and adjacent fintech verticals only works when it captures intent and context the way a loan officer would, not the way a web form does. Better's missteps — overpromising on AI replacing loan officers, then quietly walking it back — are also instructive. The carriers, brokers, and fintechs that succeed in 2026 are deploying conversational intake as a layer on top of human expertise, not as a replacement for it.

Better.com in 2026: Status, Comeback, and the Conversational Turn#

Better.com is, as of 2026, a smaller, leaner, AI-forward mortgage lender that survived one of the most public fintech implosions of the post-SPAC era. The company went public via SPAC on the Nasdaq on August 24, 2023 under ticker BETR, two years after the deal was first announced. Funded loan volume in Q3 2024 reached approximately $3.6 billion, up roughly 80% year-over-year, and the company reported a net loss narrower than analysts expected — milestones that anchored the comeback narrative.

The turnaround is inseparable from the company's bet on conversational AI. Better launched Betsy, its in-house AI voice and chat assistant for borrowers, in September 2024. By late 2025, Betsy was handling first-touch borrower conversations, pre-qualification questions, document collection prompts, and after-hours servicing inquiries that previously required a loan officer or call-center agent. The pitch from CEO Vishal Garg is straightforward: digital mortgage origination is a conversation, not a form, and the company that treats it as a conversation wins.

This is the broader shift in ai mortgage infrastructure that fintech leaders should be paying attention to. The Mortgage Bankers Association (MBA) has tracked steadily rising adoption of AI in originator workflows since 2023, and Better is the most visible public-company example of an end-to-end AI mortgage stack. For a deeper read on the #1 retail mortgage lender's parallel strategy, see Rocket Mortgage's AI playbook — Better and Rocket are now competing on conversational depth, not on rate-table SEO alone.

Betsy: What Better's AI Mortgage Assistant Actually Does#

Betsy is a multi-modal mortgage origination AI that runs the early stages of the borrower journey end-to-end. It does five things in practice:

1. First-touch qualification. When a borrower lands on Better.com from a search ad or refi calculator, Betsy initiates a conversational session — voice or chat — that captures purchase price, down payment, employment situation, and credit-band self-attestation in natural language.
2. Soft credit pull + pre-approval letter. For eligible borrowers, Betsy orchestrates a soft credit pull and returns a personalized pre-approval letter, typically in under three minutes, without handing off to a human loan officer.
3. Document collection. Betsy prompts for W-2s, pay stubs, and bank statements via uploaded files or linked accounts (via aggregators like Plaid), and explains in plain language why each document is needed.
4. Status updates + after-hours servicing. Borrowers can ask Betsy "where is my loan?" 24/7 and get a substantive answer pulled from the loan origination system, not a templated email.
5. Outbound follow-up. In 2025, Better extended Betsy to outbound voice — calling back leads who abandoned an application, asking what blocked them, and re-engaging them conversationally.

The architectural point worth highlighting: Betsy is not a chatbot bolted to a legacy form. It is a conversational front-end designed to replace the form. That distinction matters because most fintech AI projects ship a chat widget that sits next to a form, and the form still does the actual qualifying. Better took the harder path of rebuilding intake around conversation — which is the only architecture that captures borrower intent at the depth a human loan officer would.

How Conversational AI Rebuilt the Path to Pre-Approval#

The pre-2024 Better.com flow was best-in-class for a form-based digital mortgage but still form-based. A borrower hit a landing page, filled out 20+ fields across multiple steps, waited for a call from a loan officer, then did a soft pull. Median time-to-pre-approval was measured in hours-to-days.

The Betsy-era flow compresses this into one conversational session:

• Intent capture comes first, not last. Betsy asks "what are you trying to do?" — purchase, refi, cash-out, HELOC — before asking for personal data. That single reordering is the entire form-vs-conversation insight in miniature; it's also why automated lead qualification software built around forms keeps producing junk-quality leads. Better's own published funnel data suggests conversational intake meaningfully increases completion versus the prior form-based flow, though the company has not disclosed a specific lift figure publicly.
• Follow-ups happen in-line. When a borrower says "I'm self-employed," Betsy doesn't add a checkbox — it asks how income is reported, whether the borrower has two years of returns, and whether business and personal accounts are commingled. This is the same probing logic a senior loan officer uses, and it surfaces deal-killers early.
• Disclosures happen contextually. Regulatory disclosures (TILA, RESPA, GFE) are delivered when the borrower asks a question that triggers them, not as a wall of text at the start.
• Soft pull triggers automatically. With consent captured conversationally, Betsy initiates the soft credit pull as soon as it has enough data to underwrite a pre-approval letter.
• Hand-off is graceful, not forced. Borrowers who want a human loan officer can request one mid-conversation; Betsy passes the full transcript so the loan officer doesn't restart.

For fintechs running similar intake-heavy funnels, the broader pattern is documented in the conversational intake playbook for high-friction intake forms and the conversion gap between forms and conversations hit 4x in 2026 trend analysis.

Lessons from Better's Missteps: AI Hype vs. AI Reality#

An honest case study of Better.com requires honest accounting of what went wrong, because the missteps are as instructive as the wins. Three matter for fintechs evaluating AI mortgage intake today.

1. The 2021 SPAC and the layoff Zoom call. In December 2021, Better laid off roughly 900 employees on a single nine-minute Zoom call — about 9% of staff — in an incident widely covered by The New York Times, Forbes, and CNBC. Internal morale collapsed, several senior executives resigned, and the company entered its SPAC year with a reputation crisis it has spent four years rebuilding. SEC filings from the period detail subsequent restructurings and the SPAC delay.

2. The AI-will-replace-loan-officers narrative. Better's early Betsy marketing leaned hard on the implication that AI could replace loan officers. The company has quietly walked this back: Betsy now handles the front of the funnel and routes complex cases to humans. The framing matters because it's the same mistake Klarna made and reversed with its 700-agent replacement claim — and the same mistake every "AI replaces X" pitch invites. The correct frame, which Better has gradually adopted, is that AI handles the conversational intake layer; humans handle the high-stakes underwriting and closing conversations.

3. Rate-environment whiplash. Better's business model in 2020–2021 was optimized for ultra-low rates and a refi boom. When rates rose, the business model didn't translate. AI intake is not a substitute for a defensible rate strategy or a balanced purchase/refi mix — and any fintech intake ai roadmap that ignores macro is mispriced.

The honest read: Better is succeeding at conversational intake while still working through the structural issues of being a mortgage lender in a high-rate environment. For fintechs adopting similar architecture, the right takeaway is that conversational AI is a force multiplier, not a turnaround strategy. Adjacent fintech case studies worth studying include DocuSign's path from forms to conversations, Stripe's onboarding philosophy, and Ramp's customer onboarding strategy.

What Intake-Heavy Fintechs Should Learn from Better#

Most fintechs that operate automated lead qualification software today — mortgage, personal lending, insurance, small-business loans, BaaS providers — face a version of the same problem Better had in 2023: their intake forms convert at single-digit percentages, leads abandon at the credit-pull step, and loan officers (or relationship managers) spend half their day re-qualifying leads the form already failed to qualify.

Here is what the Better playbook actually generalizes to:

• Replace the form, don't decorate it. Adding a chat widget next to a 30-field application doesn't fix the form. The conversational layer needs to BE the application. The architectural test is documented in most AI-native onboarding tools aren't native — here's the real test.
• Capture intent before identity. Ask what the borrower is trying to do before asking who they are. This is the single highest-leverage change most fintechs can make. The pattern is detailed in form abandonment is a CFO problem in 2026.
• Treat outbound follow-up as a conversation, not a drip. Better's 2025 expansion of Betsy to outbound voice is the move most fintechs are still missing. An abandoned application is a conversation the borrower wants to finish; a drip email isn't a conversation. See the MQLs are dead, conversational qualified leads are the new lead currency analysis.
• Don't overpromise AI-replaces-humans. The post-Klarna, post-Better consensus is that AI handles the conversational intake layer and humans handle high-stakes closing. Anyone selling you "fire your loan officers" in 2026 is selling 2022's pitch.
• Build for verticalized regulation. Mortgage has TILA, RESPA, and ECOA; auto lending has Reg Z; insurance has 50 state DOIs. Conversational intake systems that ignore this fail their first compliance review. The insurance state-of-the-industry report walks through the parallel pattern in carriers.

The Perspective AI Angle: Conversational Intake Without Building a Betsy#

Most fintechs are not going to build Betsy. They don't have Better's engineering org, training data, or willingness to absorb the failure modes of a custom-built mortgage AI. What they need is the conversational-intake layer as a productized capability — the intake conversation surface that captures intent, qualifies leads, collects documents, and hands off to human underwriters without the multi-year build.

This is the lane Perspective AI is built for. Our Concierge agent is the conversational front-end that replaces a static form — it asks questions in natural language, follows up on vague answers, captures context a form can't, and qualifies the lead before it hits the loan officer's queue. Our Interviewer agent is the research layer for fintechs that want to learn why borrowers abandon at specific funnel stages — what most lenders need is not another analytics dashboard but actual conversational data from the borrowers who churned out.

The market structure is straightforward: Better, Rocket, and a handful of large lenders will build their own. Everyone else — community banks, credit unions, fintechs at the $500M-to-$5B origination band, and adjacent verticals like insurance MGAs and personal lenders — needs a productized conversational intake layer. Perspective AI is built for CX and product teams running modern intake without a 50-person AI engineering build-out.

Run a Perspective study against your own funnel — start with the borrowers who abandoned your application in the last 30 days. The conversational data is more informative in two weeks than the form-funnel analytics will be in a year.

Frequently Asked Questions#

What is Better.com Betsy?#

Betsy is Better.com's AI mortgage assistant, launched in September 2024. It handles first-touch borrower qualification, soft credit pulls, pre-approval letter generation, document collection, status updates, and outbound follow-up — all through conversational voice or chat. Eligible borrowers can receive a pre-approval letter in under three minutes without speaking to a human loan officer. Betsy was expanded in 2025 to support Spanish and outbound voice follow-up for abandoned applications.

Is Better.com profitable in 2026?#

Better.com was not yet consistently profitable as of late 2025, but the company narrowed its net loss meaningfully in 2024, with Q3 2024 funded volume of approximately $3.6 billion and roughly 80% year-over-year growth. The AI-driven cost structure — particularly Betsy's role in reducing per-loan support and qualification costs — is central to management's path-to-profitability narrative. Investors should consult Better.com's latest SEC filings (investor relations) for current financials.

How does Better.com's AI compare to Rocket Mortgage's AI?#

Better.com's Betsy and Rocket Mortgage's AI stack solve overlapping problems but with different architectures. Rocket has the larger origination volume and a longer-tenured AI program embedded across Rocket Logic, while Better went further on a fully conversational front-end that replaces the application form rather than supplementing it. Both companies now compete on conversational depth rather than rate-table SEO. For a fuller comparison, see the Rocket Mortgage AI strategy case study linked above.

What is conversational mortgage origination?#

Conversational mortgage origination is the loan origination workflow in which the borrower's intent capture, qualification, document collection, and status updates happen through natural-language conversation — voice or chat — rather than through a multi-page web form. The borrower describes what they're trying to do in their own words; the AI follows up, surfaces deal-killers, and orchestrates the underlying loan origination system (LOS). It's the architectural successor to form-based digital mortgage applications.

What can other fintechs learn from Better.com's AI strategy?#

Other fintechs can learn three things from Better.com. First, replace the form rather than decorating it with chat — conversational intake only works if it becomes the application. Second, capture intent before identity; ask what the borrower wants to do before asking who they are. Third, don't overpromise AI replacing humans — Better walked back its early "Betsy replaces loan officers" framing, and the durable model is AI for the conversational intake layer plus humans for high-stakes closing.

Is automated lead qualification software a good fit for mortgage lenders?#

Automated lead qualification software is a strong fit for mortgage lenders, but only when it is built around conversation rather than form fields. Forms in mortgage produce single-digit completion rates and junk-quality leads; conversational qualification — the architecture Better.com deployed with Betsy — captures borrower intent, surfaces self-employment or credit complications early, and produces qualified leads that loan officers can actually close. The wrong implementation is a chatbot bolted to a 30-field form; the right one is a conversational front-end that IS the application.

Conclusion#

Better.com's comeback is, at its core, a story about replacing forms with conversations — and about being honest with the market about what AI can and can't do in mortgage. Betsy isn't a miracle; it's a productized version of what good loan officers have always done, applied at the volume only software can deliver. The lesson for the broader fintech-intake market is that automated lead qualification software in 2026 is a conversational-intake category, not a form-optimization category. The lenders, insurers, and adjacent fintechs that internalize this will compound a conversion advantage that compounds every quarter. The ones that don't will keep A/B-testing field counts.

For fintech and CX teams that want the Betsy-style conversational intake experience without building it from scratch, start a study with Perspective AI or explore the Concierge agent for form-replacement intake. The conversational layer is the durable advantage; the form is the legacy.