Photoroom launches Visual Agents to fix flaws in product images, so companies don't need to build their own AI infrastructure

Photoroom announces Visual Agents, a managed system that performs AI-powered image and video production for enterprise product catalogs. The launch is accompanied by internal benchmark results that show that the best AI image editing models on the market pass a full product fidelity check just 29% of the time and distort or render logos unreadable in one out of every five results. Photoroom tested 4,250 AI-generated images on 850 real products, covering clothing, shoes, bags, jewelry and accessories. Photoroom's Fidelity Layer, which Visual Agents runs at every stage, raised this index to 38.2%, the best result in the test.

This need is why several of the world's largest marketplaces have built their own internal image scoring systems. Until now, there was no commercial option.

The challenge for companies in visual production with AI is no longer capacity, it is integration. Teams need to figure out which tools solve their visual problems and then stitch together multiple models and vendors to cover what a single tool can't do alone. What's left of this are inconsistent results, quality review that remains manual, and custom workflows that cost more to maintain than they cost to build.

Visual Agents eliminates this work. The system reads each image received, selects the most suitable models for it, evaluates the final result in relation to the product it was supposed to show and automatically repeats any failed attempt, forwarding only genuinely borderline cases to a human reviewer.

In testing, Photoroom engineers found that the same item of clothing could fail a fidelity check and pass the next attempt, without any changes to the model in between. Automatic retry recovers some of these failures before a reviewer sees them, so human review is reserved for cases that really need it.

Early enterprise implementations include:

• One of the world's largest consumer marketplaces, which ran more than 1.5 million automated quality checks in three weeks and corrected more than 90,000 images where a check found a defect;
• A food delivery marketplace that processed 8,200 catalog images in a single overnight run, with 99.9% completed automatically and seven forwarded to a human reviewer;
• A global fashion resale marketplace that measured a 2.2% increase in gross merchandise volume over seven days compared to a control group, ranging from 2% to 5% by product category, using Create.

“I've run imaging operations that processed millions of products a year, and the part that no one sees is the review queue. Each image that needs a person represents a minute of someone's day, and on the scale of a catalog, those minutes turn into headcount. Visual Agents reduces that queue to the images that actually need someone's intervention,” says Jeff Strauss, Head of Imaging at Photoroom, previously at Gap, Amazon and Wolt.

“Almost every corporate team we talk to describes the same problem. They have good models at their disposal, and yet at the end of the process they can't get a catalog they trust. This isn't a problem with the model, but with everything around it, like knowing which model to use for each product, checking if the result is good enough to sell, and fixing it when it isn't. This work had been falling into our customers” laps, so we built Visual Agents so that it falls into ours instead," he explains. Matt Rouif, CEO and co-founder of Photoroom.

Visual Agents is now available to enterprise customers, with volume-based pricing. Companies with more stringent loyalty requirements can add the Business Guarantee, the contractual commitment to the quality of results. The agent first runs its own retry cycle, and the guarantee covers what still fails to meet the standard. Details in Visual Agents.