Background
Most AI tools send your data to a data centre and return an answer. Colibri does the opposite: the AI is installed on your computer and the data stays where it is. This page explains what that means in practice, and why we built it this way.
When you install Colibri, it downloads a set of 18 small AI models to your machine. Each one does a single job well: describing what is in an image, recognising faces and text, transcribing speech and telling speakers apart, translating, summarising, finding similar content. Together they read every photo, video, audio file and document in your archive, and they do it on your own processor and graphics card.
After that first download, Colibri works without an internet connection. There is no account, no API key and no server on our side that ever sees a file. If you unplug the network cable, the search still works.
Publishers hold unpublished manuscripts, licensed artwork, interviews with named people and contracts. Archives hold material whose rights holders never agreed to a cloud upload. With Colibri, this question does not arise, because the material never leaves your systems. There is no transfer to a third country, no data-processing agreement for a foreign server and no vendor who reads along.
Colibri does not train any model with your files. What it produces are descriptions, transcripts and search indexes in a Colibri database on your machine, optionally also as sidecar files next to your data. Delete them, and nothing remains. That is what the GDPR calls the right to erasure, and here it is a single click rather than a request to a provider.
The same applies to copyright. Style guides, licensed characters and photos from agencies stay inside the licence you already have, because they are never copied anywhere.
Large cloud models answer every request with hundreds of billions of parameters, in data centres that need cooling water and a power supply the size of a town. Small, specialised models need a fraction of that. A laptop can run them, and it only runs them once: each file is analysed a single time, and every later search or question costs almost nothing.
This is also why Colibri runs on hardware you already own. A current Mac or a Windows PC with a graphics card is enough for everyday work. Nothing has to be shipped, replaced or rented.
Cloud AI is priced per request, per file or per month, and the bill grows with the archive. Colibri is a licence you buy once. There is no cost per query, no cost per terabyte and no subscription that switches the archive off when it lapses.
Independence goes further than price. Colibri does not depend on a single AI provider's interface, so a change in someone else's terms cannot change your workflow. Your files stay in your folders in their original formats, the metadata lives in a Colibri database next to them and can also be written out as open sidecar files, and your existing systems keep running. Colibri is a layer on top, not a replacement, and you can remove the layer at any time.
Only once, to download the AI models after installation, and later for updates. Analysis, search and the question-and-answer dialogue run entirely offline.
No. Colibri does not train any model with your files. It only creates metadata and search indexes on your computer, and both can be deleted completely at any time.
For searching and browsing an analysed collection, a Mac with Apple Silicon (macOS 14 or newer) and 16 GB of memory, or a Windows 10/11 PC with four cores and 16 GB of RAM, is enough. Importing and analysing large archives is faster with 32 GB or more and a dedicated graphics card; for production pace we recommend a Mac with an M-series Max chip or a Windows PC with an NVIDIA RTX card.
Colibri uses 18 small models, each specialised in one task such as scene description, speech recognition or translation. For these tasks they match large general models, while using a fraction of the energy and keeping every file in the building.
Yes. Colibri reads file servers, cloud folders and existing databases where they are and adds a knowledge layer on top. Nothing is migrated. It runs as its own app or invisibly behind the tools you already use.
See It Live
The beta is running. Bring a question about your archive, and Colibri answers it on your hardware.