“A group of young people looking at their mobile phones”

Many publishers have vast image archives—yet a large portion of them remains unused. Why? Missing or inadequate captions make images practically invisible.

And even with current content, the same question often arises: How do you find an image if what you’re looking for was never tagged?

The answer lies in modern AI.

Instead of traditional auto-tagging, we rely on multimodal LLMs and vector search. This allows editors to simply describe what they want to see—and receive relevant results from millions of images in seconds.

No metadata at all. No manual tagging at all.

🔍 Examples from our test system (approx. 4 million images)

“A diver in a coral reef”

→ Detected immediately and accurately

“A pair of elephants at the zoo”

→ Not just elephants—but specifically a pair

“A person is reading a printed newspaper in a café”

→ The more detailed the description, the better the results

“BMW Presentation”

→ Logos, colors, and fonts are reliably recognized—even without captions

All images have been indexed automatically. The search is based solely on image content—metadata is not required.

This means that your existing archives will be searchable immediately—even retroactively.

Visual Search is only the beginning.

Comyan AI Search offers six powerful ways to find and work with images — including Image Matching (to any text), Concept Search, Visual Filters, Similar Image Search, and Tolerant Name Search.

Help your editors find the right content faster — and get more done, every day.