Vision model 2.4

Zurla Vision 2.4

It looks before every ad lands.

A frontier vision model, built for advertising. It classifies every page, image, frame, screen and ad, reads the intent in every image, and keeps out what your rules don’t allow.

Book a demo See it at work

Looking at pages, frames, screens and ads

Illustrative data · fictional brands

What it does

It looks.
It labels.
It keeps every ad in the right place.

  1. Topics and context

    Pages, images, video frames and app screens are classified by topic and context. A hotel ad finds a travel guide, and a sports drink finds a morning run.

  2. Intent from the image

    It reads the intent in an image, not just what’s in it. A trail at sunrise says a run is coming; a hotel balcony says a trip is being planned. Each ad meets the moment its image points to.

  3. Brand safety

    Each brand sets its rules in plain words. Content outside them is kept out before a bid goes out.

  4. Creative review

    Every ad creative is checked before it runs: the image, the words, the page it links to and the format it was made for.

  5. Labels people can see

    It checks that native ads carry a clear label, like Sponsored, so readers can always tell an ad from the story.

At work

Say your rules.
See what it sees.

Brand rules used to be long keyword lists. Now you say them in plain words, and Zurla Vision 2.4 applies them to every page, frame, screen and creative it sees.

Zurla Vision 2.4 Calderwick
Ask Zurla about brand safety
1.4Mpages, frames and screens checked for Wrenquay Bank
884placements matched to Vantleigh Hotels’ campaign
6 of 6Velstead Motors creatives cleared, every native ad labeled

Illustrative data · fictional brands

Where it works

Every page, screen and creative,
seen before the ad runs.

It looks at whatever an ad will sit next to: a page, a video frame, an app screen, or another ad.

Safety and control

It looks at content,
not at people.

Zurla Vision 2.4 classifies what an ad would sit next to. It doesn’t need to know who is reading to do it, and you decide what it keeps out.

  • Previewed first

    New brand rules are checked against today’s pages first, so you see what they keep out before they go live.

  • You approve

    You decide what is kept out. See why any decision was made, and change a rule in plain words.

  • Your brand rules

    Each brand keeps its own rules, from topics to tone, applied before a bid goes out.

  • Privacy and consent

    Each person’s choices are read and respected: IAB TCF v2.3 in Europe and the UK, GPP for US state privacy choices.

Zurla Vision 2.4

Specs

Type
Vision model for classifying content
Sees
Pages, images, video frames, app screens and ad creatives
Gives
Topic and context labels, the intent in each image, brand-safety categories, creative checks and label checks for native ads
Speed
Labels ready inside auctions that finish in under 100 ms
Where you use it
Before every bid, on Zurla’s SDK, Prebid and RTB infrastructure, and in chat with Zurla
Works with
Zurla Swarm 2.1 and Zurla Northstar 3.5

The Zurla model family

Three models.
One Zurla.

Each model does one job, and they work as one. Zurla Vision 2.4 sees the content, Zurla Northstar 3.5 sets the course, and Zurla Swarm 2.1 talks with you and acts in every auction.

Show it your pages.

Book a demo, and see Zurla Vision 2.4 classify your own pages, apps and creatives.

Book a demo Zurla for agencies