Popsa's answer is not another cloud storage subscription. Instead, it runs hundreds of machine learning models directly on the user's smartphone, analyzing faces, places, objects, and events without ever uploading the raw images. In 2025, that approach powered $58 million in revenue and a growth rate the company calls its strongest in four years .

What Is Popsa?

Founded in 2016 by Liam Houghton, Popsa is a Certified B Corporation headquartered in Soho, London . Its mobile app uses on-device AI — what the company calls the PrivateContext layer — to analyze a user's photo library and automatically create ready-to-print photo books, calendars, prints, and personalised merchandise . The app is available in 12 languages and ships to 48–50 countries .

Popsa's founding premise was that automation, not more manual editing tools, was the way to help people turn their digital memories into physical keepsakes. As Houghton explained, "Unlike cloud-based services that upload and process your photos remotely, our models analyse everything locally. In most cases, we never see your photos unless you decide to print them" .

How On-Device AI (PrivateContext) Works

Rather than sending personal images to a remote server for analysis, Popsa compresses and deploys its machine learning models directly onto the user's iPhone or Android device . The on-device approach means:

• Face recognition and naming — The app identifies people in photos and matches them to names the user provides, all processed locally .
• Place, activity, object, and relationship detection — Models analyse the content of each image to understand context .
• Event inference — Popsa detects when a meaningful event has occurred by recognising associated objects and cross-checking with the user's calendar .
• Duplicate and near-duplicate removal — Perceptual hashing algorithms calculate binary representations of photos to spot and eliminate duplicates, keeping only the best shot .
• Aesthetic ranking and automatic curation — The system scores images for composition, clarity, and relevance, then selects the best candidates for each album .
• Captions and titles — Popsa uses generative AI, including Amazon Nova models via Bedrock, to write personalised titles and captions. In 2025 alone, the system generated over 5.5 million AI-powered titles .

Because all personal images stay on the device, only the final curated product data (not the raw photos) ever leaves the user's phone . Popsa's technology page states the company uses "hundreds of machine learning models" running locally to accomplish this .

Business Model

Popsa monetises by converting users' curated digital selections into premium printed products. The direct-to-consumer model includes:

• AI-powered photo books with automatically generated layouts and titles
• Prints, canvases, mugs, and other personalised merchandise
• Year-in-review albums — over 1 million of these were AI-generated in 2025 alone

The average customer purchases at least twice, and many have ordered hundreds of times . Average order value increased 10% year-over-year in 2025 .

Key Metrics at a Glance

Growth Trajectory

Popsa's growth is a compounding story. The company made roughly $100k in revenue in its first full year. By the 2025 Black Friday–Cyber Monday week, it was making $100k every 30 minutes . November 2025 was the company's best month ever, both in revenue and new customers .

Other 2025 milestones include:

• First $1M+ revenue days in GBP, USD, and EUR
• Highest rate of growth in four years
• Organic search visibility up 1,300%
• Half a billion emails sent (in 12 languages)
• 17,000 video ads created (up from 7,000)

The company's revenue per employee — over $1M in mid-2025, roughly 7x the UK tech average — reflects a lean, highly automated operation .

Future Outlook

Popsa expects $70M–$80M in revenue for 2026, with the US market as a key growth driver . The company has appointed corporate PR partner Element Communications to build awareness of the "memory curation" category . Its stated mission is to evolve from selling printed products toward a broader platform for "healthier forms of reminiscence and meaning-making" — helping people identify meaningful moments and better understand themselves .

Key risks and uncertainties: The company is still at Series A stage (last round 2022), and its current valuation is unstated. Revenue per employee ($1M) is outstanding but may indicate scaling constraints ahead. The printed-photo-product market also faces long-term competition from digital-only memory solutions and large incumbents (Google Photos, Apple Photos) that also deploy on-device AI curation.

The Bottom Line

While much of the AI industry focuses on generating new images, Popsa has spent a decade doing the opposite: making sense of the real photos people have already taken . By running hundreds of models on-device, the company solves two problems at once — photo overload and privacy concerns — and has built a $58M business in the process.

Popsa's answer is not another cloud storage subscription. Instead, it runs hundreds of machine learning models directly on the user's smartphone, analyzing faces, places, objects, and events without ever uploading the raw images. In 2025, that approach powered $58 million in revenue and a growth rate the company calls its strongest in four years .

What Is Popsa?

Founded in 2016 by Liam Houghton, Popsa is a Certified B Corporation headquartered in Soho, London . Its mobile app uses on-device AI — what the company calls the PrivateContext layer — to analyze a user's photo library and automatically create ready-to-print photo books, calendars, prints, and personalised merchandise . The app is available in 12 languages and ships to 48–50 countries .

Popsa's founding premise was that automation, not more manual editing tools, was the way to help people turn their digital memories into physical keepsakes. As Houghton explained, "Unlike cloud-based services that upload and process your photos remotely, our models analyse everything locally. In most cases, we never see your photos unless you decide to print them" .

How On-Device AI (PrivateContext) Works

Rather than sending personal images to a remote server for analysis, Popsa compresses and deploys its machine learning models directly onto the user's iPhone or Android device . The on-device approach means:

• Face recognition and naming — The app identifies people in photos and matches them to names the user provides, all processed locally .
• Place, activity, object, and relationship detection — Models analyse the content of each image to understand context .
• Event inference — Popsa detects when a meaningful event has occurred by recognising associated objects and cross-checking with the user's calendar .
• Duplicate and near-duplicate removal — Perceptual hashing algorithms calculate binary representations of photos to spot and eliminate duplicates, keeping only the best shot .
• Aesthetic ranking and automatic curation — The system scores images for composition, clarity, and relevance, then selects the best candidates for each album .
• Captions and titles — Popsa uses generative AI, including Amazon Nova models via Bedrock, to write personalised titles and captions. In 2025 alone, the system generated over 5.5 million AI-powered titles .

Because all personal images stay on the device, only the final curated product data (not the raw photos) ever leaves the user's phone . Popsa's technology page states the company uses "hundreds of machine learning models" running locally to accomplish this .

Business Model

Popsa monetises by converting users' curated digital selections into premium printed products. The direct-to-consumer model includes:

• AI-powered photo books with automatically generated layouts and titles
• Prints, canvases, mugs, and other personalised merchandise
• Year-in-review albums — over 1 million of these were AI-generated in 2025 alone

The average customer purchases at least twice, and many have ordered hundreds of times . Average order value increased 10% year-over-year in 2025 .

Key Metrics at a Glance

Growth Trajectory

Popsa's growth is a compounding story. The company made roughly $100k in revenue in its first full year. By the 2025 Black Friday–Cyber Monday week, it was making $100k every 30 minutes . November 2025 was the company's best month ever, both in revenue and new customers .

Other 2025 milestones include:

• First $1M+ revenue days in GBP, USD, and EUR
• Highest rate of growth in four years
• Organic search visibility up 1,300%
• Half a billion emails sent (in 12 languages)
• 17,000 video ads created (up from 7,000)

The company's revenue per employee — over $1M in mid-2025, roughly 7x the UK tech average — reflects a lean, highly automated operation .

Future Outlook

Popsa expects $70M–$80M in revenue for 2026, with the US market as a key growth driver . The company has appointed corporate PR partner Element Communications to build awareness of the "memory curation" category . Its stated mission is to evolve from selling printed products toward a broader platform for "healthier forms of reminiscence and meaning-making" — helping people identify meaningful moments and better understand themselves .

Key risks and uncertainties: The company is still at Series A stage (last round 2022), and its current valuation is unstated. Revenue per employee ($1M) is outstanding but may indicate scaling constraints ahead. The printed-photo-product market also faces long-term competition from digital-only memory solutions and large incumbents (Google Photos, Apple Photos) that also deploy on-device AI curation.

The Bottom Line

While much of the AI industry focuses on generating new images, Popsa has spent a decade doing the opposite: making sense of the real photos people have already taken . By running hundreds of models on-device, the company solves two problems at once — photo overload and privacy concerns — and has built a $58M business in the process.