Semantic search has arrived in Granit

Type "sunset over the sea" into Granit and the right photo surfaces — no tags needed. Semantic search is now available to everyone.

Multimodal semantic search in Granit: find a photo by describing it in natural language

"Sunny beach," "the xxx brand logo" — you type that into your Granit catalog, and the right photo or media surfaces, even though you never tagged it, never described it. Semantic search is now open to every Granit user.

In short, Granit now understands what your media actually contains and lets you find it in your own words. That's what semantic search is: you search by meaning, not by labels.

Search for a photo or a file the way you'd describe it out loud

Multimodal semantic search analyzes the visual content of each media file and matches it to your query, written in plain natural language.

So you can type, in plain words:

  • "Sunset over the sea, CineStill"
  • "Black and white portrait"
  • "Outdoor wedding atmosphere"
  • "Red car against an urban backdrop"

...and watch the right media surface — no tags, no dedicated folder, no hand-written description. You can even get specific: an object, a mood and a style in the same sentence, and Granit keeps up.

What it changes when you make a living from your images

For a photographer preparing a client selection, or a studio juggling thousands of assets, this isn't a gimmick — it's a category shift. You move from manual organization — folders, tags and a lot of memory — to a catalog that understands what it holds. The time you spent filing and digging, you get back to create.

  • Find any media file in a few words, even the ones you never got around to tagging.
  • Explore your catalog by theme or by mood, not just by collection.
  • Run a "find me similar photos" in one click from any media detail page.
  • Automatically group visually consistent images to build a moodboard or a client selection. (module in progress, available late 2026)
  • Search your photos as well as the frames pulled from your videos — the technology is multimodal, and we'll add more formats over time.

Your media, always respected

Your images stay yours: they're analyzed to help you find them, not to feed an outside AI.

Frequently asked questions

What is semantic search in Granit?

It's a search that finds your media based on their visual content rather than their tags or file names. You describe what you're looking for in natural language, and Granit shows you the matching images.

Do you have to tag your photos for it to work?

No. That's the whole point: it works even on media that was never tagged or sorted.

Does it work on videos?

Search covers your photos and the frames extracted from your videos. More formats will follow, since the technology is multimodal.

Are my images used to train an AI?

No. Granit is hosted in Europe and doesn't use your work to train third-party models. The solution we rely on does not perform AI training on your media.

Semantic search is available right now.
We keep building Granit with you.