Models

AI models where they demonstrably save time.

We do not use AI for the sake of AI. Self-trained models, lightning-fast decision models, generative imagery and language models: applied where they measurably take work off your hands.

01

Self-trained vision models

For cropping, background removal and colour correction, we train our own models on product photos. More than 590,000 photos have been automatically cropped and published this way.

02

Decision models in milliseconds

Our own decision models make decisions within milliseconds in the processing pipeline. Fast enough to process thousands of products a day without any waiting time.

03

Generative imagery

Photopilot creates AI model photos and campaign images, including with a fixed house model (Model Swap).

04

Language models

SEO product copy in any language in Photopilot, an assistant with sources in Activaflow, and smart suggestions for every time entry in Half5.

05

Measurable

At one Photopilot customer, new stock is online in less than 12 hours. Results like that are what we steer by.

FAQ

Frequently asked questions

Which AI models do you use?

We choose per task. For image editing such as cropping and colour correction, we train our own models, and our own decision models decide within milliseconds. For text and generative imagery, we use the best available model, and we switch when a better one comes along.

Is AI output always correct?

No, which is why we build in checks. Photopilot has an optional approval step, and in Activaflow AI never changes the dossier on its own.

Self-made & independent

Built on our own terms.

webRidge has been self-built since 2011, without loans or external investors. We own our platforms and choose the long term: not absurd margins, but lasting relationships with our customers.