About PicsTag

Who is behind this

PicsTag is built and maintained by BeFunnyStudio, a small independent studio. It is not a funded startup and there is no sales department to get past — the tool was published because the problem it solves turned out to be everyone's problem. Emailcontact@picstag.app and it lands in a real inbox.

Everything in the guides is written by the same people who wrote the tool. That comes with an obvious bias, so it is stated plainly: we make an image tagging tool, and we think AI drafting plus human review beats both hand-writing everything and trusting a model blindly. Where the guides give advice that argues against using PicsTag — and several of them do — that is deliberate.

Why this exists

Image libraries fail for a dull reason: nobody fills in the metadata. Tagging is slow, repetitive work that pays off only after every image has been done, so it gets postponed forever and the library becomes a folder you scroll through.

AI is good at the boring half of that job — recognising what is in a picture and proposing words for it. It is not good at the half that matters, which is deciding whether those words are the right ones foryour catalogue. PicsTag splits the work along that line: the model proposes, you dispose.

Why it runs in your browser

Every comparable tool asks you to upload your images to a server. For unreleased product shots, client work under NDA, or anything covered by a data processing agreement, that upload is the whole problem.

So PicsTag does not have one. The models run in a Web Worker inside your tab, using WebGPU where it is available and WebAssembly where it is not. The site is a static page with no backend: there is no server that could receive your images, which is a stronger guarantee than any promise written on this page — and one you can check yourself in about ten seconds with your browser's Network tab.

Under the hood

• Tags come from an image classification model (ViT), which returns its top predictions with a confidence score each.
• Captions come from an image-to-text model (ViT + GPT-2).
• Both run through transformers.js and are cached by your browser after the first download.
• An optional cloud mode can call a more capable vision model through OpenRouter, using an API key you provide and that never leaves your browser.

What it deliberately does not do

Saying what a tool is not is more useful than another feature list, and it saves you finding out the hard way:

• It does not write metadata back into your image files. It produces a CSV or JSON sidecar that you import. Embedding IPTC into a JPEG in the browser is possible but easy to get subtly wrong, and a corrupted master file is a worse outcome than an extra import step.
• It does not store anything. No accounts, no history, no cloud copy of your queue. Reload the tab and it is gone. That is the cost of having no server, and it is why the tool nags you to export.
• It does not claim the output is finished work. The local models are small ones chosen to fit down a normal internet connection. They are a good first draft and a poor final answer — seehow accurate AI image tagging actually is, which is blunt about where they fail.

How it is paid for

The tool is free and has no usage cap, because the expensive part — running the model — happens on your machine rather than on a server someone has to pay for. The running costs are a domain and static hosting. The site carries advertising on its written pages to cover that; the tool itself stays free either way.

Contact

Feedback, bugs, or a feature you need: contact@picstag.app. Recent changes are listed in the changelog.