Customer: Cardless

Cardless helps consumer brands build credit products into their own digital experiences. As the company grew, it needed to strengthen identity confidence and fraud detection while keeping onboarding fast and easy for legitimate applicants.

We spoke with Gabe Kahn, Credit Risk Lead at Cardless, about how the company balances fraud prevention with conversion, how Plaid Identity Verification fits into its risk-based onboarding flow, and how the company is preparing for the next generation of identity fraud.

Plaid: To start, what does Cardless make possible for brands and their customers?

Kahn: Cardless builds embedded credit infrastructure that lets consumer brands launch and operate credit card programs directly inside their apps and websites. Through a single API layer, we handle issuance, underwriting, servicing, compliance, and capital markets, so brands can own the customer relationship. For brands, that can mean more applications, stronger conversion, and deeper loyalty. For customers, it means a credit product within a brand they already trust, with relevant rewards.

Today, we power programs across crypto, travel, and retail.

Plaid: Those programs serve very different audiences. How does that shape your approach to identity and fraud?

Kahn: Every brand brings a different customer base and risk profile. A crypto audience looks very different from an airline or rent-payment audience, so we can’t rely on a one-size-fits-all identity stack. 

Every extra verification step also creates another opportunity for a legitimate applicant to drop out. We need stronger scrutiny where the risk is real, while keeping the experience nearly invisible everywhere else. That balance has become harder as generative AI makes it easier to bypass methods that were previously reliable, pushing more cases to manual review and adding friction for customers and strain for our team.

Plaid: Is that where your previous approach was falling short?

Kahn: Yes. Having the wrong verification solutions can be the worst of both worlds: lost good customers on one side and missed fraud on the other.

Plaid: What made Plaid stand out, and how do you use it today?

Kahn: We focused on decision accuracy, an experience customers would actually complete, and the ability to catch fraud our previous stack may have been missing. Plaid Identity Verification declined fewer legitimate applicants, while its full liveness check surfaced new risks.

We use it as a risk-based step-up rather than routing every applicant through the same process. When we detect signals associated with identity theft or first-party fraud, we route that applicant through enhanced verifications. Everyone else stays on a low-friction path.

Plaid: What changed once that approach was in place?

Kahn: Stronger verification now lands exactly where it’s needed and stays out of the way everywhere else. The full liveness check has also helped us identify difficult-to-catch fraud. A lighter check could miss those situations.

Plaid: What has that meant for conversion?

Kahn: Identity verification completion—the share of applicants who reach the step-up and finish verification—increased significantly after we moved to Plaid. Because that step-up sits at a high-intent but historically high-drop-off point, improving completion at that stage contributes directly to more completed applications and approved accounts. 

Plaid: Has there been an impact on operational efficiency as well?

Kahn: Absolutely. Fewer false positives mean fewer legitimate applicants enter manual-review or support queues. That lightens the workload for our fraud operations and review teams and lets them focus on genuinely ambiguous cases. Plaid also makes it easier to spot anomalies, saving time on each review.

Plaid: How does that flexibility support partners?

Kahn: The risk-based model lets us adjust identity scrutiny for each program without rebuilding the entire application flow. 

Plaid is also straightforward to implement in partner front ends, which matters as we onboard new programs. Its product team has worked directly with us and our data science team to explore product opportunities, troubleshoot issues, and listen to our feedback, while its account team has helped us explore capabilities for other areas of the business, like pre-qualification solutions.

Plaid: That kind of collaboration seems especially important as fraud evolves. What concerns you most looking ahead?

Kahn: Synthetic identity fraud and generative AI. Fabricated or blended personally identifiable information that can pass traditional bureau checks is becoming cheaper and more convincing at scale. Generative AI is also reducing the cost of document forgeries, deepfake selfies designed to defeat liveness checks, and scripted social-engineering attacks.

Signals that once seemed conclusive on their own are becoming easier to spoof in isolation. Durable defense requires layering independent, difficult-to-fake signals rather than relying on any single check. That’s where we see Plaid becoming even more valuable: combining document and liveness verification that continues to improve against deepfakes with bank-linked and network-level signals. Staying closely aligned will help us keep evolving our defenses as fraud techniques change.