Articles

Pipe CEO Claurelle Rakipovic sits down with TechEdge at Money2020 2026

At Money20/20 2026 in Las Vegas, Pipe CEO Claurelle Rakipovic sat down with TechEdge to talk about where AI belongs in fintech, why a high compliance bar can be an advantage, and how embedded capital creates value for small businesses and the software platforms they run on.

Highlights

AI and people split the work by problem type. Rakipovic separates knowledge problems, where machines are faster, from judgment problems like policy and exceptions, which stay with people.
Compliance is a design input. She compared AI agents to the early days of card payments: adoption followed once protections were built in.
Proof of value shows up on the P&L. Pipe measures AI on financial results, customer experience and system security.
Underwriting runs on operational data. Pipe reads the data that already lives on a partner's platform, so merchants can access capital where they already work.
Built for new markets from day one. Pipe has expanded into the UK, Canada, Australia and New Zealand over the past two years, and taken as little as 3 weeks to launch in each geo.

Watch the full interview

The full 16-minute conversation with Jarrett Banks, Editor-at-Large at ExecEdge's Tech Edge, is on ExecEdge: Fintech for SMEs: Pipe CEO Claurelle Rakipovic, live from Money20/20 2026.

How does Pipe decide what AI handles and what people handle?

Rakipovic draws the line between knowledge problems and judgment problems. Machines take on the knowledge load, while people define policy, exceptions and the outcomes the business is solving for.

A fraud review is a knowledge problem: a person might weigh seven pieces of information, while a machine can pull them together and reach a conclusion much faster. Pipe now runs a fraud agent on top of that work. Its credit decision models also replaced what used to be a multi-step human process.

Judgment stays with people. That includes setting policy, deciding exceptions, and choosing which data an agent sees versus a person.

"AI has made it easy in some ways to be done, but not necessarily with the standards or the measures that we think matter." (3:11)

For more on how Pipe pairs technology with people, read Embedded capital support: how blending tech-first and a human touch drives utilization and customer satisfaction.

Does fintech's compliance bar make the industry cautious about AI?

There is caution, Rakipovic said, but the industry has been here before. When consumers moved from cash to cards, protections like refunds for disputed purchases made adoption possible.

She sees the same pattern with AI agents. Fintechs know their space well enough to build the protections that let new technology work safely.

"What really needs to be true is that you are using that not as a blocker, but using the advantage you have as a fintech." (4:46)

Internally, she pointed to large language models turning compliance policies into technology requirements, actions and owners. Externally, agents will need authentication and permissions, and the protocols for that are already emerging.

What does proof of value look like for AI at Pipe?

At Pipe, AI has to prove itself on the P&L, in the customer experience and in the security of the system. Rakipovic said that discipline also drives the decision between building and buying.

One example came from collections, where an agent prepares the packages that help merchants going through a hard stretch get a different plan.

"We have converted that into now almost 20 to 30 cases that can be worked fully." (7:45, up from around five per hour)

The clearest proof, she said, is underwriting.

"Our credit underwriting models are our first example of an industry making space for this technology and turning it into really deep penetration of businesses that never would get access to capital if we did not exist." (8:40)

How does embedded capital create a flywheel for small businesses and platforms?

Pipe funds businesses based on how they actually operate, those businesses grow, and the platforms they run on grow with them. Rakipovic called it a simple flywheel.

Instead of generic P&L or bank data, Pipe looks at operational data such as seasonality. That lets Pipe meet a restaurant on Uber Eats or a med spa on Boulevard where it already works, and offer capital without an application.

Owners use it for new salon chairs, a new location, an oven or payroll for an expansion.

"Those things might sound small, but they create real economic value, and for the platform, it creates more turnover for them overall." (10:48)

She sees a bigger flywheel ahead as small businesses come to expect software that anticipates what they need, including capital at the moment they need it.

How did Pipe expand into the UK, Canada, Australia and New Zealand so quickly?

Pipe’s platform was set up to handle global expansion from day one, so each one took weeks to stand up. Rakipovic said a comparable international expansion took six years at a previous company.

The difference was starting with extensibility in mind. Money moves between merchants, Pipe, funding providers and countries, so the rails have to be designed around edge cases from the start, such as how a partner's accounting ledger differs across five countries.

"You actually have to start with, this is not a roadmap. This is a Lego block." (14:28)

Pipe's underwriting models were built the same way: relatively agnostic to industry and tuned to currency across regions. That approach let Pipe launch the UK in under six weeks.