How LendingClub's AI Agent Strategy Reshapes Bank Verification for MCA Lenders

Borrowers Are Finding Lenders Through AI Agents, Not Google

During LendingClub's Q1 2026 earnings call, stock analyst David Scharf asked CEO Scott Sanborn a question that should concern every MCA lender in the market: how is LendingClub preparing for a world where borrowers use AI agents, not Google, to shop for a loan? Sanborn's answer confirmed that LendingClub is actively investing in AI-type search channels and positioning its platform to be discoverable by autonomous AI agents acting on behalf of borrowers.

This is not a theoretical scenario. AI fraud detection for business lending has historically focused on catching manipulated documents, synthetic identities, and doctored bank statements. But the emergence of AI agents as loan shopping intermediaries introduces a fundamentally different problem. When an AI agent submits an application on behalf of a borrower, lenders lose visibility into whether the borrower actually initiated the request, whether the financial data was accurately represented, and whether the entire interaction is legitimate. For MCA funders and brokers who rely on speed to fund, this shift demands a rethink of how verification workflows operate from the first touchpoint to the final funding decision.

The New Fraud Surface: AI-Intermediated Applications

How AI Agents Generate and Submit Loan Applications

The mechanics are straightforward. A small business owner tells an AI assistant, whether it's ChatGPT, a Perplexity agent, or a purpose-built fintech bot, to find the best merchant cash advance for their situation. The agent scrapes available data, fills out applications, and submits them to multiple funders simultaneously. In some cases, the agent may even negotiate terms or respond to follow-up questions autonomously.

For the lender on the receiving end, these applications look clean. The data is well-formatted. The responses are coherent. But the underlying problem is that the lender has no way to verify whether the human behind the AI agent actually authorized the application, whether the financial details were accurately sourced, or whether the agent introduced errors or fabrications during the process. This is especially concerning given that LLM referrals are already reshaping MCA lead quality and fraud risk in measurable ways.

Why API-Only Verification Falls Short Against Agent-Originated Fraud

Many lenders rely on API-based bank connectivity tools that pull transaction data directly from financial institutions. These tools work well for confirming account ownership and pulling structured transaction histories. They do not, however, verify that the person who initiated the data-sharing flow is the same person who owns the business and authorized the advance. An AI agent can, in theory, use stored credentials or pre-authorized tokens to initiate a bank data pull without the merchant's real-time involvement.

The gap becomes more dangerous when you consider that static bank statement PDFs are already trivially easy to forge using generative AI tools. As we explored in our analysis of how auto lending fraud tactics are migrating to MCA bank verification, sophisticated fraud rings are applying techniques honed in the auto lending space to merchant cash advance applications. AI-intermediated applications amplify this risk by removing the human interaction that underwriters traditionally rely on to detect inconsistencies.

What Layered Verification Actually Looks Like

Effective AI fraud detection for business lending in 2026 requires more than a single verification method. The most resilient approach combines three layers. First, document-level verification confirms that bank statements and financial records are authentic and unaltered. Second, behavioral verification confirms that a real human performed the banking actions being reviewed. Third, audit trail integrity ensures that every step of the verification process is timestamped, stored, and reviewable.

Asynchronous screen recording addresses the second and third layers in ways that neither live calls nor API pulls can. When a merchant records their live banking session through a browser-based tool, the resulting video captures mouse movements, page load sequences, URL bar contents, and real-time data rendering. This creates evidence that is extraordinarily difficult to synthesize. Unlike a static PDF or an API data dump, a screen recording of a live banking portal session shows the data in its native context, with all the visual cues that an underwriter needs to assess authenticity.

Exact Balance was built around this principle. Applicants receive a secure link, record their banking portal at their convenience using browser-based screen capture with no software installation, and submit the recording for review. An AI-guided floating coach walks them through each step and verifies completion in real time. The result is a full audit trail with timestamped activity tracking that documents when links were opened, recordings started, and submissions completed.

What This Means for MCA Funders and Brokers on the Ground

Consider a scenario that is becoming increasingly common. A funder receives three applications in a single morning for different businesses, all submitted through the same AI-powered lending marketplace. The applications are well-structured, the bank statements look clean, and the financials pass automated screening. But the funder has no way to confirm that three separate business owners actually sat down and authorized those applications. Without a verification step that requires human presence, the funder is exposed to coordinated application fraud at a scale that manual review cannot catch.

This is not a hypothetical risk. The Consumer Financial Protection Bureau has repeatedly flagged the risks of automated lending processes that lack adequate human verification checkpoints. While MCAs have been excluded from the CFPB's small business data collection requirements, as confirmed in the agency's final rules filed on April 30th, the regulatory direction is clear: lenders are expected to demonstrate that their verification processes are robust enough to prevent fraud, regardless of whether the application arrived through a human broker or an AI agent.

For funders processing high volumes, the operational challenge is real. You cannot schedule a live verification call for every deal without destroying your turnaround time. But you also cannot skip verification when the application channel itself may be compromised. Asynchronous verification solves this tension. The merchant records on their schedule, the underwriter reviews on theirs, and the entire process produces a compliance-ready audit trail without a single scheduling conflict.

Brokers face a parallel challenge. As AI agents begin submitting applications on behalf of merchants, brokers need to demonstrate to their funder partners that the deals they bring are legitimate. A video recording of a live banking session, attached to every deal package, is a powerful trust signal. It separates high-quality brokers from those whose pipelines may be contaminated by agent-originated fraud. We previously examined how trusted data channels prevent fraud in MCA broker-to-funder workflows, and the AI agent era makes those channels even more critical.

Frequently Asked Questions

What are AI agents in lending and how do they affect MCA applications?

AI agents in lending are autonomous software tools that act on behalf of borrowers to search for, compare, and sometimes submit loan or MCA applications. They affect MCA applications by removing the direct human interaction that underwriters rely on to assess intent and authenticity. When an AI agent fills out an application and submits financial data, the lender cannot easily confirm that the business owner authorized the submission or that the data was accurately represented. This creates new fraud surfaces that require verification methods capable of confirming human presence and real-time data accuracy.

How does asynchronous bank verification detect AI-generated fraud?

Asynchronous bank verification detects AI-generated fraud by requiring the applicant to perform a live, recorded session in their actual banking portal. The screen recording captures real-time page loads, dynamic data rendering, URL navigation, and mouse movements that are extremely difficult for AI to fabricate convincingly. Unlike static documents or API data pulls, a video of a live banking session provides visual and behavioral evidence of human presence. Platforms like Exact Balance add AI-guided step verification and timestamped activity logs to create a comprehensive, tamper-resistant audit trail.

Do MCA lenders need to change their verification process because of AI agents?

Yes. MCA lenders who rely exclusively on document review or API-based bank data pulls should add a verification layer that confirms human involvement in the application process. AI agents can submit well-formatted, internally consistent applications that pass automated screening but lack genuine borrower authorization. Adding asynchronous screen recording to the verification workflow provides the behavioral evidence needed to distinguish legitimate applications from agent-originated or synthetic submissions, without slowing down deal velocity.

Is video-based bank verification compliant with lending regulations?

Video-based bank verification supports compliance by producing timestamped, encrypted recordings that serve as audit-ready evidence of the verification process. Each recording documents exactly what the applicant showed, when they showed it, and what steps they completed. This is more thorough than a live call with no recording or a static document review with no chain of custody. While specific compliance requirements vary by jurisdiction, the audit trail produced by platforms like Exact Balance meets the documentation standards that regulators and auditors expect.

Conclusion

LendingClub's public investment in AI agent channels is not an isolated experiment. It signals a structural shift in how borrowers find and apply for funding, one that will ripple through the MCA industry faster than most funders expect. The lenders who thrive will be those who recognize that verification must evolve alongside origination. When applications can arrive through autonomous intermediaries, the only reliable way to confirm authenticity is to require the borrower to demonstrate their banking data in a live, recorded session.

Exact Balance provides exactly this capability: asynchronous, browser-based screen recordings with AI-guided step verification and full audit trail documentation. No scheduling, no software installs, no gaps in compliance. Visit exactbalance.ca to see how async bank verification fits into your workflow and prepares your operation for the AI-intermediated deal flow that is already here.