Popsa says on-device AI helped push revenue to $58 million

London-based Popsa says it generated $58 million in revenue last year by applying AI to a less fashionable part of the image market: organizing existing smartphone photo libraries and turning selected images into printed products. The company expects revenue of $70 million to $80 million this year, with the U.S. among its fastest-growing markets and on track to become its largest.

The company did not disclose its valuation, total funding or current headcount. It says it operates in more than 50 countries and produces about $1.2 million in revenue per employee, a metric it attributes partly to using AI across internal operations as well as in its consumer product.

Popsa’s pitch sits apart from the current rush to generate synthetic images and video. The company uses AI to assess and organize photos people have already taken, then convert them into photo books, calendars and other printed formats. According to research by Popsa, people take an average of 551 photos a month, or more than 6,600 a year, while 70% of photos are stored and not revisited.

On-device photo analysis

Co-founder and CEO Liam Houghton said Popsa was shaped by his own experience managing a personal archive of more than 200,000 images, including digitized childhood photos, older computer files and printed photographs with added dates and locations.

Popsa says its app analyzes photos for people, places, activities and relationships, then removes duplicates and selects images for albums. The company says the app generates about 89,000 smart albums, though it did not specify the period attached to that figure.

A central product decision, according to Houghton, was to run analysis on the user’s device rather than sending photo libraries to the cloud. He said Popsa generally does not see a customer’s images unless the customer chooses to print them.

The company says hundreds of models are involved. Some score image quality, including composition and lighting. Others identify activities, recurring people and the difference between someone central to a user’s life and someone captured incidentally in the background. Popsa’s broader claim is that its systems infer meaning from patterns in a library rather than just tagging objects inside a single image.

From curation to captions

Popsa is also developing automated storytelling features. The company says its AI converts images into descriptive data that captures locations, relationships, activities and emotions, then uses that information to create captions and narratives. Popsa says 12 million captions were generated in the last year based on photo data.

The company is positioning that work as memory preservation rather than general photo printing. Houghton said Popsa is not trying to become a seller of mugs or keyrings, and that its focus is on helping users create records that can be shared across generations.

Popsa’s research found that 77% of Europeans have not planned what should happen to their digital photo collections after they die. The company says that gap supports demand for printed annual photo books and other physical archives, although it did not provide customer retention or repeat-purchase figures.

Profitable growth without a new raise

Popsa says it has been EBITDA-profitable since 2022 while continuing to increase annual R&D spending. Houghton said the company has grown without needing additional capital and may raise funding in the future, but wants to retain the option not to do so.

Internationalization has been part of the product from early on, according to the company, including support for multiple languages, currencies, local holidays and date formats. Popsa also produces its advertising internally and expects to create about 30,000 video ads this year across its markets, using AI for personas, dubbing and localized creative variations.

As AI-generated media spreads, Houghton argues that verified personal photographs will become more valuable as records of real events. Popsa allows users to upload AI-generated images, he said, but the company is betting its larger opportunity is in making overloaded camera rolls usable.

This story draws on original reporting from Tech.eu.