Andrew Bud and the trouble with being sure

Andrew Bud had already built a business that moved money when he discovered how little movement could prove. A payment could travel through the system and still lack the one thing that mattered: a person who had genuinely agreed to it. In 2008, fraud at mBlox, the mobile business he had founded, made that distinction painfully public.

Bud has returned to the episode in interviews rather than tidying it out of his biography. He describes a forensic investigation conducted by employees over a long weekend, and a lesson about hearing warnings before events make them impossible to ignore. The company’s response, he says, was subsequently praised by the regulator. The experience nevertheless changed the direction of his work.

By 2011, he had founded iProov. The question underneath it was disarmingly small: could an ordinary device establish that an actual person was physically present? A career spent helping machines communicate had arrived at the business of checking who, exactly, was communicating.

A career spent making things connect

Bud’s professional identity begins with engineering. He studied at Christ’s College, Cambridge, between 1979 and 1982, earning a first-class engineering degree. His student activities included co-founding and editing Energy Matters, a publication concerned with energy policy and technology. Even the undergraduate version seems to have wanted a technical problem and somewhere to argue about it.

He describes a London upbringing in a family of entrepreneurs and industrialists whose businesses included electrical fittings, batteries and scientific instruments. Engineering carried family prestige. His mother had earned a doctorate in chemistry in post-war Britain; his father was creative and continued working into his nineties. Bud’s account gives ambition a domestic setting, with accomplished relatives providing both encouragement and something to live up to.

His early work included fusion technology at Culham and MIT. He loved the subject, but its pace frustrated him. Microprocessors offered a faster-moving field, and he entered it through PA Technology. In his telling, the changes of direction followed opportunities to make something happen within a working lifetime.

Digital wireless telephones followed, then wireless data and mobile networks in Italy. At Olivetti, and through work associated with Omnitel Pronto Italia, he took part in the shift toward mobile communications. The geography matters: his career developed across British and Italian organisations, with technical work gradually giving way to responsibility for businesses.

His contribution to DECT, the digital cordless telecommunications standard, involved network protocols and data capabilities. Standards rarely produce a satisfying photograph. Their achievement is that different pieces of equipment can agree on what happens next. That quietly demanding work belongs in the story of an entrepreneur whose later company would also depend on getting distant parties to agree.

A little light, a difficult proof

Bud founded mBlox in 1999, according to iProov’s account of his career. It grew into a business supplying enterprise text messaging. The attraction was scale: a service that could handle large volumes of communications for other organisations. The fraud episode later revealed the consequences of relying on an inadequate authentication mechanism.

The first iProov problem was technical enough to resist a good sales pitch. A picture resembling a person could be copied. A recording could be played again. A remote service needed evidence about the encounter happening now. Meanwhile, the user might have little patience for instructions and little interest in buying special equipment.

The company’s approach uses the device screen to illuminate a face with an unpredictable sequence of colours. Video travels back for analysis of the reflected light. The relationship between the challenge and the returning signal helps establish that the capture is fresh and involves a physically present person. It turns a familiar screen into part of the test.

An early government innovation directory catches the less polished work behind the proposition. In 2013, iProov was studying data from smartphone sensors for an experimental biometric project. The team found inaccuracies that needed correction and collected enough information to make identifying useful patterns a task in itself. The next step was still a prototype.

That record is useful because it restores uncertainty to the beginning. A promising idea had to become measurable evidence. Data collection, correction and feasibility work occupied the space between an invention and a product someone else could reasonably depend on. The prototype did not arrive with its eventual biography attached.

A towel, a daughter and an awkward demonstration

Bud’s account of an early bank evaluation contains a more memorable test. iProov was using another vendor’s face verification technology. A bank executive asked his daughter to try it. The system confused her with her father, ending the commercial prospect.

A family resemblance had become a business problem. Bud says the failure helped prompt a proposal from iProov’s research director to develop a recogniser internally, supported by Innovate UK funding. It is an awkward anecdote to volunteer when your business concerns identity. It also explains a consequential technical decision with considerably more force than a product roadmap.

For entrepreneurial nerve, Bud reaches for Douglas Adams. He likes the towel in The Hitchhiker’s Guide to the Galaxy, particularly its usefulness against a creature whose understanding of visibility leaves something to be desired.

“Every entrepreneur needs that towel.”

Andrew Bud

The joke allows for a little necessary ignorance at the start of a venture. An exhaustive appreciation of everything that might go wrong could prevent anybody from beginning. The bank story supplies the accompanying obligation: once a test reveals a problem, the result deserves your attention. Optimism can open the laboratory. It cannot grade its own experiments.

The years before the numbers

iProov’s public history separates its founding in 2011 from its commercial launch in 2013. That distinction gives the early work room to exist. Bud remembers the first Innovate UK grant as a significant moment for a tiny team trying to make an unproven technology into a business.

Other people helped carry the idea. Joe Palmer joined in 2011 as employee number two and built the Flashmark prototype. Dominic Forrest, who had worked at mBlox, joined in 2013. Andrew Newell arrived from University College London in 2014 and led image-processing work. The company’s technical history contains several Andrews, which presumably rewards careful meeting invitations.

The scale became visible in the figures iProov published around its January 2022 investment. Revenue had tripled between 2020 and 2021. During one ten-day stretch in 2021, it processed more verifications than during all of 2020. Several days exceeded a million checks. These are company-reported figures from that period, rather than a description of today’s daily traffic.

Sumeru’s investment brought resources for expansion and managing partner Kyle Ryland onto the board. Institutions using the company’s technology included the UK Home Office, the US Department of Homeland Security, GovTech Singapore, ING and Rabobank. The engineering question had travelled into settings where a failed identity check could interrupt banking, government administration or a journey.

Bud later explained the timing of the funding in practical terms: raising money before it became necessary preserved choices. Capital could buy room to act. For a company defending against changing attacks, that room had an operational purpose beyond the pleasure of announcing a round.

The letter that made his hands shake

In November 2019, a different kind of validation arrived from the Cabinet Office. The letter concerned Bud’s appointment as a CBE in the 2020 New Year Honours, for services to exports in science and technology. His wife, he recalled, noticed his hands shaking as he read it.

The investiture eventually took place at St James’s Palace. Princess Anne asked about his path from engineer to entrepreneur and about deepfakes. Bud enjoyed the building and the conversation. He was especially moved by his children’s response to the honour: work they had watched him pursue had become publicly legible.

“I’ve been a struggling entrepreneur most of my life.”

Andrew Bud

In 2020 he also became a Fellow of the Royal Academy of Engineering. The two recognitions illuminate different parts of the same career: the technical contribution and the business of taking it into the world. His own account keeps the uncertainty in view, even after the letters have accumulated after his name.

Who said yes to the machine?

By June 2026, Bud was discussing human presence in an environment increasingly populated by AI agents. At Identity Week Europe, he argued that services would need ways to connect automated activity to accountable people. An agent could perform work; consequential decisions would still raise questions about the person behind it.

In September, iProov published the Human Approval and Presence Specification, or HAPS. The experimental framework addresses a particular gap: access credentials alone do not show that a person approved a specific proposed action. It invites organisations to request evidence of human approval at selected sensitive moments.

The proposal leaves room for different ways of establishing presence. Its public repository describes an experimental draft and partial reference implementation, with limitations that require further work. That status is part of the substance. A specification offered for examination needs to survive examination before it can justify confidence.

Bud’s newer work returns to the old payment problem with more capable software involved. A system can possess permission and still need evidence of a person’s intention. The question that redirected his career has acquired fresh company: when the machine says yes, whose agreement does that represent?