France and the US are the two countries which have shaped the career of Matthieu Rouif (known as “Matt”), the co-founder and CEO of Photoroom, one of the most talked-about startups in the European AI firmament in recent times.
Born in France, the Y-Combinator and Stanford University alum spent time in the French Air Force (over 100 parachute jumps to his name) before a career which has seen him yo-yo across the Atlantic, by way of teaching jobs, founding startups, or working at US tech firms.
Likewise, France and the US are the two markets which have been central to propelling the Photoroom app to 200m-plus downloads and Hollywood recognition: the Paris-headquartered startup has a soon-to-open office in the US (New York), its biggest market, where it plucks talent from while also counting some US heavyweight brands as customers.
“I think people are very smart on both sides of the pond,” says Rouif, speaking over video from the US.
In a breezy 30-minute video interview, Rouif, who is in his late thirties, discusses Trump tariffs; a typical day in the office; his relationship with his co-founder; European sovereignty and Project Europe; the DE&I backlash; banning direct messaging in the workplace; spending millions of Euros on AI training data; and why he quit skydiving.
Photoroom’s raison d’être
Founded in 2019, Photoroom’s raison d’être is democratising photo editing: it’s an easy-to-use photo editing app which does a good job of matching the quality of a professional photographer minus the lofty expense.
Photoroom is primarily known for removing background imagery and creating a white background for e-commerce sellers (its core market), and casual users.
It has also expanded its offer to AI tools, so users can swap a rainy Paris background for sunny Bali or remove an unwanted facial pimple while users can also edit photos en masse. It has free offerings and subscription packages, up to $90 a year.
Occupying a highly congested market, Photoroom competes against the likes of Adobe, Google, Canva, Piscart and a slew of other AI startups.
But Photoroom appears to be motoring along: it processes around five billion images a year, hit €50m in annual recurring revenue in 2023, is profitable, and has recently shifted into the enterprise sector, luring in Amazon and DoorDash as customers.
It also won a role in the Barbie film, as Warner Bros leveraged its photo editing API in the film’s social media campaign.
Some numbers
Last year, Photoroom raised $43m at a $500m valuation, in a round led by Balderton Capital, with Aglaé and previous backer Y Combinator also participating.
Other backers include Meta and Meta AI head Yann LeCun and execs from Hugging Face and Disney+. It has raised a total of $64bn.
Photoroom didn’t magic itself out of thin air, Aladdin lamp-like, as Rouif had been working in and around photo technology for around 15 years.
He got the “itch” to launch another startup while working as a senior product manager at GoPro, which acquired Stupeflix, the previous startup he was working at.
He teamed up with co-founder, Eliot Andres, Photoroom’s CTO, a former child actor, whose resume includes published tech author and speedy app developer.
The pair, Rouif reflects, are “quite complementary” in style.
Typical day
So what is a typical day for Rouif? After the school run ( he has three children), Rouif says his days are, effectively, split into two, as he presides over a team of around 80.
He says mornings are spent listening to its internal podcast with user interviews, strategy work, and product development, while afternoons are spent in internal and external meetings.
He says he enjoys the public-facing part of the job, like speaking to the media. That said, he says he wants to get Andres to do more of it.
Attracting talent
Amid an insanely competitive machine learning talent market, Rouif says that a pull for potential recruits is the freedom Photoroom offers builders.
He says: “Machine learners and engineers come to Photoroom and a few months later they see a model they built from scratch reach 100m people. You are not going to be able to do that at bigger companies.”
He also points out that, amid a whirlwind of AI startups cropping up, Photoroom is one of the few which is financially secure.
Furthermore, he says Photoroom has a “no direct” Slack message policy (inspired by Stripe’s anyone-can-read-an-email policy), an edict which, he says, gives engineers the chance to jump on any projects they like should they have a lightbulb moment.
Training models
Photoroom trains its own image AI models, with its training data sourced from publicly available data, images submitted by users, as well as data purchased from professional photographers and photography firms.
In total, he says Photoroom has spent millions of Euros on buying and licencing training data.
By comparison, OpenAI, behind ChatGPT, has spent hundreds of millions of dollars licencing content from news publishers to train its AI models.
Along with ethical concerns about crediting and paying for training data, another big ethical concern with AI is deepfakes and the harm they can cause.
Rouif says: “As we train our models from scratch, we control the data. We can use data that is clean, so we know where it comes from. It is a bit like when you buy food, it has traceability.
“The value of creating from scratch is we don’t show sex or violence to our models because we don’t need it for e-commerce.”
He also points out that Photoroom’s biggest use case is the background image, while copyright laws in many countries are limited to the main subject.
Trump tariffs and DE&I
Amid the ongoing hoo-ha and fallout following Trump’s tariffs, Rouif says the tariffs have had “no impact” on Photoroom” but points out it’s an ever-changing landscape.
He says: “The GPUs are already rented and already distributed on servers. And we also actually train in Europe with clean energy.”
The Trump government has also swept in a backlash against DE&I policies, with Amazon and Meta heel-turning on such policies.
Rouif has previously posted on social media about Photoroom hiring specifically for a female for a role at the startup.
Clarifying this, he says Photroom’s hiring approach is “meritocratic” but says, statistically, men apply for more roles than women for roles at Photoroom, so sourcing talent is important to improve diversity.
European sovereignty and Project Europe
Another rumbling debate is European sovereignty and Rouif says he is “pumped” to be part of Project Europe, Harry Stebbings’ fund aimed at investing in the next generation of European founders.
Rouif says Photoroom was built with the idea that Europe could build a global leader, with talent from around the world.
Riffing on this, he says role models, who have made it in the US, are key to helping European startups.
He also stresses the importance of getting going at an early age (Rouif launched his first startup at Stanford when he was 22).
He says: “When I came back from Stanford, I was in France and I tried to bring all the entrepreneurs to talk to students to show them it’s possible.”
Photoroom’s future
These days, post fatherhood, Rouif has called time on the skydiving. But he still has big ambitions for Photoroom, as he looks to see off competition from AI startups, which have a magpie eye for mimicking rivals.
On Photoroom’s agenda for the rest of 2025 is the release of new models, while it’s also looking to beef up its client base.
And then, as the interview draws to a close, Rouif utters a sentence which would likely sound mawkish coming from a non-Parisian. He says he wants to “make the internet a bit more beautiful with beautiful images”.