Payment Compliance, Fraud Prevention, and the Rise of Agentic Commerce

Ryft CEO Sadra Hosseini joined Channel X TV to discuss why payment compliance is now a central decision for platforms, how fraud detection has evolved for complex multi-party transactions, and what agentic AI means for the future of commerce.

‍

From hospitality marketplace to payment service provider

Sadra is a third-time founder. His second company was a hospitality marketplace platform. One app, all venues, where customers could order food and drink without waiting at the bar or flagging down a waiter.

"We started pre-Covid, Covid hit, and we scaled rapidly to about a million customers. It was acquired about three and a half years ago."

Whilst building that business, Sadra saw where commerce was heading. Volume was shifting from legacy processors to digital platforms and marketplaces, which now have 85% market penetration with SMEs. Platforms wanted to bring payments in-house, improve the user journey, and monetise transactions in a compliant way.

That insight became Ryft, a payment system built for platforms, marketplaces, and any digital business managing complex payment flows.

"On every hundred million processed, platforms make an extra £1.27 million going through their platform."

‍

The problem with legacy payment providers

Most legacy processors offer marketplaces something called a merchant of record. The processor settles all funds into a single account, and the platform is then responsible for paying out to its sellers or third parties.

This creates two serious problems.

Problem one: it is non-compliant. A regulation came into effect in December 2020 stating that marketplaces cannot hold funds unless they hold a payment licence. Obtaining that licence takes up to 18 months, and only 5% of businesses succeed. Most platforms simply cannot meet that requirement.

Problem two: it creates operational overhead. The platform must manually calculate and execute payouts to each party. That introduces human error, delayed settlements, and the need to build an internal balance system tracking who is owed what across every transaction.

The alternative is to use a payment provider that handles the compliance layer and automates the split at source.

"If you use the right payment provider, they can have that fully integrated into the marketplace itself."

With Ryft, each transaction is split automatically via API. The marketplace receives its fees, each seller or recipient receives their share, and the balance is held by Ryft rather than the platform. No manual reconciliation. No compliance exposure.

‍

Payment data as a growth tool

Complex payment flows generate rich data. With the right infrastructure, that data feeds directly into a platform's analytics.

Sadra explains that businesses working with Ryft can pull transaction data through the API into their own analytics tools. This goes beyond knowing which products sell. It tells you:

• Which geolocations drive the highest conversion rates
• Where acceptance rates drop and why (for example, a UK product sold to a German buyer)
• Which product categories drive the highest margin
• Which seller or supplier segments are most valuable

"It's not just purely oh, product X does really well. It's also where should you be selling that, which geolocations, and what kind of people should you be targeting."

This level of insight is only available when the payment provider is deeply integrated into the platform, not sitting at arm's length as a legacy settlement account.

‍

Commerce 2.0: when AI agents do the buying

In Commerce 1.0, the buyer visits a website, browses, adds to a basket, and checks out. The journey is human-driven. In Commerce 2.0, the buyer tells an AI agent what they want. The agent searches, compares, selects, and completes the purchase without the buyer visiting a website at all.

"Payments are moving from the open internet, where you go to a website and buy some shoes for yourself, to the closed internet, where you tell your AI to go and buy the shoes for you."

This is already happening. ChatGPT Operator, available on the Pro plan, can open a browser, find a product, and complete a purchase autonomously.

What does this mean for platforms and merchants?

Businesses that currently optimise for human buyers, through Google Shopping, pay-per-click, and product imagery, will need to rethink their approach. AI agents do not browse. They parse data. They respond to structured, context-rich product information: specifications, availability, pricing, location, delivery windows.

"Platforms and digital businesses will need to learn: how do I start marketing to the agents themselves rather than to a human being?"

Businesses built on infrastructure that controls the payment flow and owns the data will be better placed to adapt. Those relying on legacy processors will not have the visibility or flexibility to move quickly.

‍

Fraud prevention in complex payment flows

Digital commerce fraud is harder to manage when transactions involve multiple parties. A single payment may flow between a buyer, multiple sellers or suppliers, a shipping provider, and a platform. Any one leg could be fraudulent.

Sadra outlines how Ryft handles this:

Contextual data. When a dispute arises, Ryft passes the platform all available transaction context, including buyer location, purchase destination, shipping address, and account history. The platform can use this data to make an informed decision, or delegate that decision to Ryft.

Split-level blocking. If a transaction is split five ways and one leg appears fraudulent, Ryft can block that leg specifically, reverse it, and return funds to the buyer without disrupting the legitimate parts of the transaction.

Machine learning at scale. Ryft's fraud detection runs through a machine learning algorithm that improves continuously through feedback loops. Every confirmed fraudulent transaction is fed back into the model, increasing accuracy over time.

There are three categories of fraud that any payment system managing complex flows must handle:

• Simple fraud: clearly suspicious signals such as mismatched billing and shipping countries
• Friendly fraud: genuine buyers with legitimate but unusual behaviour (buying internationally, using freight forwarders)
• Full fraud: deliberate financial crime with significant risk exposure

"The right product will be able to pick out each one of these, put a risk score against it, and then block the wrong transactions."

Sadra acknowledges the nuance required. Treating every unusual transaction as fraud damages legitimate customer relationships. Treating full fraud as friendly fraud creates financial exposure. The machine learning model is built to distinguish between the two, and to improve over time.

‍

Why sellers benefit from FCA licensing

The PSD2 regulation that created the compliance problem for platforms also created a protection mechanism for sellers.

Under the master merchant model, all funds flow through the platform's own account before being paid out to sellers. If the platform becomes insolvent, those funds are at risk. Sellers have no direct claim on money sitting in a third-party account.

When a platform works with an FCA-licensed provider like Ryft, the payment flow changes entirely. Ryft holds the funds, splits the transaction, and pays each party directly. The platform never holds seller funds.

"If the marketplace goes under tomorrow, there is no risk for the sellers. They will still get their money out."

Seller onboarding speed

Legacy processors require two to three weeks of documentation, questions, and manual verification before a seller can begin trading. With Ryft, seller onboarding takes approximately five minutes and sellers can go live the same day. For anyone listing a single product or testing a new channel, a two-week process is not viable. Fast onboarding reduces drop-off and directly increases active seller count.

‍

Why choose Ryft as your payment service provider

Ryft is an FCA-licensed payment service provider  built for platforms, marketplaces, and digital businesses managing complex payment flows. Businesses using Ryft can accept payments online, automate split payments to unlimited sellers, onboard sellers in minutes, and access full transaction data through a single API integration.

‍