AI Visual Search

Search it the way you'd say it

Type "close-up of the packaging, plain background" and there it is. Tagbox.io searches what is inside the photo, the filename, the description, the tags on it and the words spoken in your videos - all from one search box.

woman holding a red bag outdoors
womanred bagoutdoors
3 results · 0.4s
Woman walking outdoors holding a red bag✓ match
Free text search

You remember the photo, not the file name

Search what you remember. The color, roughly what was happening - the search works with that. It will even work for that photo from two years ago you already gave up on.

Woman on a street holding a dark red handbag
AI Image Search Tagbox
the one with the red bag, outdoors
Photos and video

Land on the second, not the file

Analyzed every second and transcribed. One search covers your photos and your footage.

Shopper facing a wall of sneakers beside a stacked tower of shoeboxes
the shoebox tower wobbling01:38
Text in image

The label on the box is searchable

Tagbox.io reads the words inside the photo - the brand, the line, the size. All of it searchable the moment the file lands.

Marlo Oat Crisp cereal box beside a bowl of oats
Read from the photo
“Marlo”“Oat Crisp”“Gluten Free”
Multilingual platform

100+ languages. Search in yours.

Search, transcripts and the interface each follow the person, so a team spread across three countries works natively and still shares one library.

Search in your language

"Chaussures rouges sur l'herbe" brings back the red shoes on the grass.

Transcribed and translated

A panel recorded in Spanish is transcribed in Spanish, and translated if you need it.

The interface too

Each person sets the app's interface to the language they think in.

Search plus your tags

The AI does not know everything

It reads the picture. It cannot know the client, the campaign or what you are cleared to run - so your tags do the rest.

Client
Bay Hotel GroupNorthline+13
Venue
Pier 27The Armory+9
Campaign
Spring 26Launch+21
Usage rights
Paid socialWeb only+4

Stop hunting. Start finding.

Frequently
asked questions

"Is this keyword search over file names and tags?"

No. It reads the picture. Describe what you remember - "red sneakers on a wet street", "the wide shot before the doors opened" - and it returns frames that match the meaning, not files whose names happen to contain those words. Synonyms, near-misses and roundabout descriptions all land.

"So what can it find with nothing tagged?"

Objects, activities, settings and colors; any text visible in the frame; the people you have named; and anything said out loud in a video. From the first upload, on every plan.

"What can it not find?"

Anything that is not visible or audible - which client paid for the shoot, which campaign it ran in, when the usage rights run out, who held the camera. That is what tags carry, and they filter the same search.

"Which parts of this are on which plan?"

AI Visual Search, text-in-image, reverse image search, similar images, color and people filters are on every plan from Starter. Logo detection and Video analysis Pro start on Pro.

"We shoot forty near-identical product variants. Can it tell them apart?"

Generic AI cannot - it will call all forty "a running shoe". Teaching it "Vertex Runner, coral" is Custom AI Tagging, which starts on Pro with five brand-aware tags.