The pitch sounds almost too straightforward: What if the real problem with AI agents spending money isn't teaching them how to pay, but deciding whether they should be allowed to?
That's the bet Ralio, a London-based fintech, is making with a €2.1 million pre-seed round announced this week. Sure Valley Ventures led the financing, which the company says closed three times oversubscribed within months of incorporation. The round also drew a surprisingly long tail of backers—Seed X out of Liechtenstein, Love Ventures, Plug and Play, Rule30, Adeline Arts and Science, Endurance Ventures, Campus Fund, and fintech angel Alan Morgan, among others. Antler, an early supporter, stayed in.
At its core, Ralio is building what CEO Ghali Bennani Laafiret calls a "safety and authorization layer" between autonomous AI agents and the payment systems they're increasingly expected to navigate. Think of it less as plumbing for agent transactions and more as a bouncer checking IDs before anyone gets near the cash register.
The Checkpoint Model
Most infrastructure being built for AI commerce focuses on enabling transactions—connecting agents to APIs, simplifying payment flows, teaching machines to handle invoices. Ralio is trying something adjacent. The platform sits as a mandatory checkpoint before an agent's payment instruction executes, enforcing programmable guardrails: spend limits, approval workflows, identity verification for the agent itself. Everything gets logged in what the company describes as an immutable audit trail, mapped back to the specific agent behavior that triggered the transaction.
It's a subtle but important distinction. Where competitors might help an agent figure out what it needs to buy, Ralio decides whether that purchase should happen at all—then documents the decision before money moves.
On the rails themselves, Ralio's website lists support for UK and European bank transfers (Faster Payments, BACS, SEPA), major card networks, and stablecoins including EURC, USDT, and USDC. There's also mention of a "Ralio Wallet" for instant settlement. Worth noting: the funding announcement doesn't include public confirmations from any of these payment partners, and the company hasn't yet named live customer integrations.
The Team Behind It
Bennani Laafiret previously built out the banking and fintech division at papernest, the French consumer services aggregator, before moving into revenue operations roles at health insurer Alan and workplace software company LumApps. His co-founder, CTO Leonardo Rosales, comes from Form3, a cloud-native payments platform where he worked on infrastructure for real-time transactions.
In the announcement, Bennani Laafiret framed AI agents as "autonomous economic actors"—a phrase that sounds grandiose until you consider that some enterprises are already letting automated systems negotiate contracts, approve vendor payments, and rebalance ad spend without human signoff. His argument is that existing payment infrastructure wasn't designed for non-human decision-makers, which creates compliance gaps, fraud exposure, and a general lack of accountability when things go sideways.
One Antler partner, explaining the fund's early backing, pointed to "moving value safely" as one of the defining challenges in AI adoption over the next decade. Perhaps. Or perhaps it's just smart positioning in a market where the regulatory frameworks haven't caught up to the technology yet.
A Nascent, Crowded Space
Ralio enters a market that barely existed 18 months ago and is now suddenly crowded. Natural, a U.S.-based competitor, raised $9.8 million last October. Paid, focused on results-based billing infrastructure, pulled in $21.6 million in September. Google introduced something called the Agent Payments Protocol—an open standard meant to streamline transactions initiated by AI—around the same time, with early adoption from Klarna and identity verification provider Trulioo.
But the space is messy. Security researchers flagged vulnerabilities in agent-led payment flows earlier this year, including prompt-injection attacks and context-binding issues that could theoretically let bad actors manipulate transaction logic. That might explain why Ralio is emphasizing the checkpoint model: when autonomous systems control financial decisions, the ability to pause, verify, and override becomes more valuable than speed alone.
The company was incorporated in England on June 27, 2025, according to Companies House records, and currently lists between two and ten employees. First accounts are due early next year. The fresh capital will fund platform development and team expansion, though Ralio hasn't disclosed how much runway the round provides or what specific milestones it's targeting.
For now, the startup is operating in a curious in-between space—too early for proven traction, but late enough that the thesis feels plausible. Whether enterprises actually need a dedicated authorization layer for AI payments, or whether that functionality just gets absorbed into existing financial infrastructure, remains an open question.
But closing a round three times oversubscribed as a company barely out of incorporation suggests investors think the answer might matter sooner than expected.
