Local AI for Media Management

Colibri is a desktop application that reads what is inside your files. It analyses photos, videos, audio recordings and documents with 18 small, specialised AI models, writes descriptions, tags, transcripts and summaries, and makes the whole archive searchable in plain language. Everything runs on your own Mac or Windows computer. No file is uploaded, no cloud account is needed, and the software keeps working without an internet connection.

A knowledge layer over the archives you already have

Colibri does not replace file servers, cloud folders, databases or archive systems. It reads them where they are, connects what it finds, and lets you search and ask across all of them. Describe a scene, look for a sentence someone said in an interview, or ask a question and get an answer with the exact source: the second in the video, the page in the document, the image.

Private by design, sustainable by architecture

Because nothing leaves the building, Colibri is GDPR-compliant without data-processing contracts for servers abroad. Nothing is used to train models, and every index can be deleted completely. Small models on normal hardware need a fraction of the energy of a large cloud model. The licence is bought once, with no cost per file or per query. Why local, actually? →

For publishers, archives, media production and everyone with too many files

Colibri is used by publishing houses, archives, film and media productions, communication teams and private collections. See the use cases, the full feature list, or apply for the beta.