The useful thing about a passport is that it settles arguments. The absurd thing is how much it says when the question is small. A bartender asks whether you are over 18; the card volunteers your name, birthday, photograph, address and a document number. Yoti was built around a neat objection: perhaps identity should arrive with an edit button.
The London company lets a person prove selected facts through a Digital ID, or pass an age check with a live facial image. In the latter case, its software estimates an age, tests that the face is real rather than a photograph, mask or replay, returns the required answer and says it deletes the image. The website may learn only “over 18.” It need not learn who you are. In an internet trained to collect first and invent a purpose later, that is a pointed bit of product design.
The founding irritationThe smallest possible answer
Yoti’s origin story begins in a queue. In 2014, co-founders Robin Tombs, Duncan Francis and Noel Hayden attended a Spartan Race where roughly 10,000 people registered using passports or driving licences. The paperwork was slow, the documents were valuable, and the process treated a physical card as the only credible bridge between person and claim. They imagined those verified details on a phone instead.
“Yoti” stands for Your Own Trusted Identity. The first public product, launched in 2017, was a free consumer app: scan a government ID, take a biometric selfie, and create reusable credentials. Attributes are encrypted and stored separately rather than kept as one fat customer record. The private key stays on the phone. When a business requests information, the person sees the request and chooses whether to share it.
A good identity product should answer the question being asked - not empty your wallet onto the counter.The design thesis behind Yoti's data minimisation
The expensive prelude£95 million before the turnstile moved
This was not a weekend wallet. By 2021, Yoti said it had spent £95 million developing the platform. Its founders and angel investors supplied most of the capital. A 2023 package added £12.5 million of HSBC debt and £7.5 million of convertible debt from existing shareholders, after Lloyds Banking Group had invested £10 million earlier that year. Yoti said total funding then exceeded £166 million.
The long build exposes what failed first: not encryption or document scanning, but the cold-start logic of reusable identity. A wallet becomes convenient only when many places accept it; merchants care only when enough people carry it. Ten million app downloads by 2021 sounded large, yet Yoti still needed bridges into ordinary transactions. Its answer was to become useful before the network became universal.
Three snapshots of a widening network
It added hosted identity verification, document checks, anti-money-laundering screening, e-signatures, authentication and developer SDKs. More important, it found a claim that businesses must check repeatedly and users would rather not document repeatedly: age. Facial age estimation does not require the consumer to create an account or present an ID. That makes it less a wallet feature than a fast lane around the wallet.
The product in one breathCapture, estimate, challenge the spoof, delete
A Yoti age-estimation check looks anticlimactic by design. The camera finds a suitable frame. A specialised model turns the face into numbers and estimates age. Liveness and anti-spoofing systems look for evidence of a replay, mask or synthetic injection. Yoti returns an age or a threshold decision. Then the image goes. The company says people demonstrating the product often respond, “Is that it?”
What happens during the one-second promise
That sequence is the company’s clearest difference from document-first rivals. Entrust’s Onfido, Jumio, Veriff, Persona, Trulioo and Sumsub all sell serious identity machinery. Yoti does too. But Yoti pairs enterprise verification with a consumer credential network and a specialist age product that can remain anonymous. The check is not facial recognition: it is not trying to discover a name or match the face against a population. It is estimating one characteristic.
The distribution machineA bank, a post office and an app walk into a bar
Yoti’s best partnerships solve the network problem rather than decorate a press release. Post Office EasyID puts Yoti technology behind a familiar British institution and connects digital checks to more than 11,000 branches for people who need help in person. Lloyds Bank Smart ID gives the same infrastructure a bank’s shop window. Digital ID Connect lets a business accept certified wallets through one integration.
Meta supplied another kind of reach. Instagram began offering Yoti’s facial estimate in 2022 when some users needed to verify their age. Yubo introduced age checks across its community. OnlyFans and Avakin Life became named customers. NCR put the technology into self-checkout software. The customer list stretches from social platforms and adult content to gaming, hiring, property, finance and retail because the underlying jobs repeat: admit the right person, keep out the wrong one, satisfy a rule, and avoid turning the sign-up screen into an interrogation room.
The business model is equally plain. The app is free for individuals. Organisations pay for checks or licences. Yoti says it does not sell personal data. This matters because the trust proposition would collapse if the company behaved like advertising technology while asking for passports and faces. Its paying customer is the gatekeeper; its product to that customer is a verified result.
The first thing to failThe birthday boundary bites
An age estimate is still an estimate. Hair, glasses, expression, lighting and the face itself can move a result. Independent NIST evaluation has shown that algorithmic performance varies by age and image set. The system is most awkward exactly where a legal threshold matters: the 17-year-old and 19-year-old do not look like separate database columns.
Yoti’s practical answer is a buffer. A retailer enforcing 18 may automatically approve only people estimated above a higher “Challenge 25” or “Challenge 30” line. Someone close to the line falls back to a Digital ID or document. That protects minors but can inconvenience young-looking adults. So what fails first is the frictionless promise, not necessarily the safety check. A cautious deployment rejects uncertainty and asks a second question.
Bias remains the issue that no cheerful demo can settle. Yoti publishes accuracy tables across age, gender and skin tone and submits models to outside testing. That is better than asking customers to take a black box on faith, but publication is not proof of equal experience in every setting. Averages can hide the people stopped at the edge. The conditions, thresholds and fallback completion rate matter as much as mean error.
The market finally arrivesRegulation is both tailwind and tripwire
What changed Yoti’s commercial gravity was not a sudden conversion inside the company. The world moved toward the problem it had been building for. The UK Online Safety Act made robust age checks urgent for services carrying adult content. Social networks faced pressure to create age-appropriate experiences. Retailers wanted automated checks at self-service tills. Reusable identity stopped sounding like a speculative wallet and started looking like compliance plumbing.
The same regulation can close a door. On September 3, 2026, Yoti said it would remove its consumer ID app from Spanish app stores while appealing a decision by Spain’s data-protection authority. The company argued that aligning with the decision would require removing biometric protections central to the app. Whatever the appeal produces, the episode is a bracing reminder: “privacy by design” is a claim made by an engineer; legal permission belongs to a regulator.
Elsewhere, Yoti has accumulated certificates meant to turn that claim into evidence: UK trust-framework roles, ISO standards, accessibility conformance, NIST age-estimation evaluation and, in 2026, an announced iBeta Level 3 result for presentation-attack detection. It also built a Guardian Council for independent ethical oversight and became one of Britain’s first B Corps. None makes a model infallible. Together they make the company unusually explicit about how it wishes to be judged.
The portable lessonCopy the question, not the camera
The most transferable part of Yoti is not its neural network. It is the discipline of reducing a transaction to the smallest verified claim. Does the nightclub need a birthday, or “18+”? Does an employer need a passport copy forever, or a certified right-to-work result? Can a returning visitor reuse a proof rather than repeat the whole ceremony? Every removed field cuts storage, breach exposure and abandonment.
Then add the less glamorous half: a conservative threshold, a fallback route, an audit trail and a deletion rule. Yoti Keys, introduced in 2025, pushes the idea further by putting an anonymous reusable age token behind a passkey on the device. The person proves age once, then participating sites can accept the token without learning an identity. It is a small credential trying to become infrastructure.
Yoti’s journey is therefore not a clean pivot from failure to triumph. It is a costly, patient narrowing. The grand wallet created technology, verified users and institutional relationships. The compact age answer found a market with immediate pain. Now the company has to keep a difficult promise: know enough to let someone through, resist the fake faces, treat the edge cases fairly - and forget the rest.