The revealing thing about a payment is how little of it you can see. A card number disappears into a checkout box. A spinner turns. An approval arrives. Behind that clean moment, credentials are stored, processors make decisions, fraud systems trade signals and merchants pay an intricate stack of tolls. Colin Luce has spent most of his career backstage, close enough to notice that one quiet technical decision can become a long commercial dependency: who holds the data?
Luce’s answer is Basis Theory, the company he co-founded in 2020 and now leads as chief executive. It offers a programmable vault for sensitive payment information. A merchant can replace a card number with a token, keep the underlying credential protected and send transactions to different providers without rebuilding the customer experience around each one. The pitch sounds like security engineering. The deeper pitch is freedom of movement.
This is how Luce tends to see payments: not as a button, but as a set of incentives disguised as plumbing. Processors want volume. Networks want acceptance. Platforms want the customer relationship. Merchants want approval rates, lower costs and the ability to change course. Tokenization began as a way to reduce the danger of storing raw credentials. In Luce’s telling, it also became a way for platforms to make departure painful.
“Each merchant and customer base is a snowflake.”Colin Luce on payment performance
His practical advice follows from that view: do not rely on a single payment service provider. A large processor may learn from billions of signals across its network, but it is still optimizing a common product for many customers. A particular merchant has its own basket sizes, geographies, risk tolerance and customer habits. A second route is not merely insurance. It can be a tool for learning what works.
The apprenticeship inside the bundle
Luce studied business administration at the University of Richmond, graduating in 2008, then moved into financial technology while the category was acquiring its modern shape. At Yodlee, an early financial-data aggregation company, he met founders building products on top of bank information that had previously been difficult to reach. The timing mattered. APIs were beginning to turn financial institutions from destinations into ingredients.
In 2014, he joined Klarna as the Swedish invoicing company prepared its American push. Buy now, pay later had not yet become a familiar checkout category. Luce has said his decision process included following respected investors: Sequoia had backed Klarna, which was enough to make the unfamiliar company worth investigating. He flew to Stockholm, met the team and signed on.
The job was an education in expansion and strain. His oldest son was around eight weeks old when he started. Luce divided his time among San Francisco, Stockholm, New York and Columbus. He later described the experience as exciting and acknowledged mistakes in Klarna’s U.S. entry. That willingness to put the wrong turns next to the wins has become part of his founder voice. He is not selling clairvoyance. Asked whether Klarna’s eventual scale had been obvious, his answer was blunt: “Definitely not.”
Roles at Uphold and Figure followed, along with consulting and advisory work for Very Good Security. By the time he reached Figure, Luce was talking with large merchants about reducing acceptance costs. The same frustration kept returning. Payments was treated as a major strategic priority, yet merchants felt locked inside systems that limited what they could change. The problem was not a lack of ambition. It was a lack of optionality.
A leap, a sticky note and a narrower company
The pandemic created an unexpected pause. Luce has written that it gave him room to decide what he wanted from his career. His wife was pregnant with their third child, another son. Starting a company in that moment was daunting, but the uncertainty also clarified the choice. On his monitor sat a sticky note carrying a line attributed to Helen Keller: “Life is either a daring adventure, or nothing at all.”
There is a photograph from the seed round that captures the non-cinematic version of the leap. A neighbor spotted Luce early one morning, outside and on the phone, negotiating financing terms. Luce later posted it for comic effect. The image fits his public persona: serious about the stakes, happy to puncture founder theater. His LinkedIn handle is “vaultdaddy,” a nickname so goofy it makes the underlying product easier to remember.
Basis Theory began with a sweeping idea. Its early site proposed a tokenization API for nearly any kind of sensitive data. The team interviewed more than 100 developers and heard that protecting, using and managing such data were all difficult. Compliance competed with product work. Encryption choices created maintenance. Audit trails were scattered. Basis Theory would abstract that burden behind APIs and developer tools.
Thousands of developers signed up. Revenue did not follow in the way the company expected. The horizontal platform was elegant and hard to explain with enough urgency. Luce narrowed the company toward payments, where the founders had experience, relationships and a customer problem attached to money already moving. The principle survived: sensitive data should remain secure and useful. The wedge became specific.
That correction is one of the more useful facts about his story. Founder mythology prefers the original flash of insight. Basis Theory’s path suggests that insight is only raw material. A market still has to recognize itself in the product. Luce learned that a broad developer audience could admire the architecture without buying it. Payment teams had a budget, a clock and a reason to care about portability today.
The agent arrives at checkout
By 2025, a new buyer appeared in every commerce forecast: the AI agent. Luce is interested, but his enthusiasm comes with a working definition. True agentic payments, to him, mean giving software a credential and authority to transact on a person’s behalf. Most current examples still end with a human approving a purchase. They are embedded checkout in a new interface, not autonomous commerce.
The distinction protects against sloppy thinking. If a person confirms the sweater, familiar fraud and chargeback systems still apply. If an agent can independently decide to buy it, questions of permission, liability, identity and limits become much harder. Luce is bullish on the longer arc and openly disappointed by some present demos, particularly the product data and discovery that precede payment. Payment is the final execution layer. A perfect vault cannot rescue a bad catalog.
His more interesting near-term examples are less photogenic than an assistant booking a holiday. Software might compare invoices, choose a payment rail or manage a repetitive business purchase under defined rules. Coding is the AI use case with obvious traction, Luce has argued, so agentic systems closest to programmable work may mature first. The point is not to make shopping look magical. It is to make delegated action controllable.
“The intent from our perspective was to give merchants a voice.”On forming the Agentic Commerce Consortium
That concern led Basis Theory to form the Agentic Commerce Consortium in September 2025. Luce believed merchants were receiving less attention than AI companies, networks and payment platforms in the rush to define new protocols. The consortium’s opening position was intentionally grounded: transaction volume was still small, consumer behavior was changing, and a contest for power was already underway.
The old question had returned with a new cast. If an AI assistant discovers a product, a commerce platform hosts the store, a processor moves the money and a merchant fulfills the order, whose customer is it? Each participant has a reason to keep a copy of the credential. Luce’s preferred architecture is an open, agnostic vault or self-custody wallet. He is realistic enough to expect duplication instead.
This is where the founder’s career forms a loop. At Yodlee he saw data become available to builders. At Klarna he watched a new payment experience enter a foreign market. At Figure he heard merchants ask for lower cost and more control. Basis Theory sits beneath those visible products, trying to make the data portable before another powerful interface closes around it.
Slope over spectacle
In October 2025, Basis Theory announced a $33 million Series B led by Costanoa Ventures, with Stage 2 Capital and Moneta VC joining earlier backers. Total reported funding reached $50 million. Luce framed the round through a company value called “Slope Matters.” Growth is meaningful relative to the speed and quality of the inputs beneath it. Capital, in this view, is useful when it increases the slope rather than decorating the chart.
The announcement also carried an unscripted family touch. Two of his sons offered their reactions after watching the company get built from its earliest days. It was a reminder that the startup timeline had been running beside another one: the eight-week-old left at home during Klarna travel, the third pregnancy during the Basis Theory leap, the children now old enough to narrate a funding milestone.
Luce’s aspiration is not difficult to state. He wants merchants to control their payment data and innovate without asking a dominant provider for permission. The execution is harder because infrastructure companies live among larger businesses with their own incentives. Neutrality has to be engineered, sold and defended.
The useful lesson is broader than tokenization. Optionality is easiest to build before it feels necessary. A backup provider matters before the outage. Portable data matters before the negotiation. Clear authority matters before an agent is trusted with a card. Luce has built a company around preparing for those moments while resisting the urge to claim they have fully arrived. In an industry addicted to the next checkout, that patience may be the point.