Meet Cardless AI, the future of AI-native financial infrastructure

Meet Cardless AI. Today we’re launching three new capabilities to ensure that our customers can offer the most advanced credit cards on the market: conversational servicing, agent-controlled spending, and Parallax, our new AI model.

People use AI every day. They ask their AI agents to make grocery lists, plan their trips, and research the best card for their wallet. They’re asking questions, delegating tasks, and increasingly - letting AI agents take action on their behalf. But their financial products haven't kept up. AI can help them decide what to buy, but when it's time to pay, they still have to do the work.

Now cards are catching up. With Cardless AI, cardholders can talk to their credit card in plain language:

“Where did I eat in Austin last year?”

“What was the French-sounding place I went to last month?”

They can even give their AI agent its own virtual card, with spending limits they control, and let it spend on their behalf.

But AI is also supercharging fraud.

AI can create fake faces, forge identity documents, and help fraudsters impersonate real people at scale. Fraudsters are getting better every day. The systems banks and legacy card issuers rely on to catch fraud are not.

So we built the Parallax AI model. Parallax connects signals across identity verification providers to catch sophisticated fraud that would otherwise go undetected.

Cardless AI makes it possible. More ways for cardholders to engage in the apps they use and spend with the agents they rely on. Better protection against fraudsters behind the scenes.

Cards that talk

Cardless customers can now offer their cardholders the ability to talk to their credit card instead of simply using it. They can ask questions and give instructions in plain language.

We call this conversational servicing. Cardholders can check a balance, make a payment, set limits on merchants or categories, and change a spending limit by asking.

They can ask “what did I spend on flights this year?” or say “freeze my card.” Or a million other questions that don’t fit into the static search bar in a legacy bank’s app.

“What was the place I went to right before getting on the Subway last Tuesday?”

“My parents are in town this weekend, make them a virtual card to enjoy the city. Let’s start with $500.”

Customers can integrate conversational servicing into their existing in-app conversational experience, or build one from scratch with Cardless’ APIs.

Cards that let AI agents spend

Cardholders can now give their AI agent its own virtual card. The cardholder sets the limits, the agent spends within them. Say “create a virtual card to buy this wool hat once it's discounted below $100,” and your agent can spend up to that amount and no more.

Our virtual cards are ready for agents to use today. Everyone has a chatbot that answers questions in their back pocket. We’re building the tools that allow agents to act on the account, make a payment, or make purchases. The cardholder decides what their agent can do with the card.

Parallax is our answer to credit card fraud in the AI era

Generative AI is making it easier than ever for fraudsters to create convincing fake identities, manipulate documents, and fool traditional identity verification systems. As these threats evolve, fraud prevention needs to evolve just as quickly.

That's why we built Parallax, a proprietary AI model that identifies AI-generated and altered images, detects synthetic and manipulated identities, and continuously adapts to new fraud techniques.

Parallax also addresses another weakness of traditional fraud prevention: most identity verification providers only see the applications that they process. Fraudsters exploit this by submitting applications across multiple providers, making coordinated attacks difficult to detect. By connecting signals across verification providers, Parallax uncovers relationships between seemingly unrelated applications and identifies fraud rings that individual providers miss.

In beta testing, approximately one-third of the fraud rings Parallax identified had activity across multiple identity verification providers. Parallax is also more effective at distinguishing fraud from legitimate activity (Parallax achieved an AUC of 0.93, compared with an industry benchmark of 0.85).

Importantly, Parallax drives these results without exposing applicants' personal data, allowing Cardless and our customers to stop fraud while protecting legitimate cardholders' privacy.

Beginning today, Cardless partners can offer these features natively within their own apps, websites, and ecosystems.