AI that runs in your browser, not on a server
These tools use real machine-learning models — for summarization, sentiment analysis, named-entity recognition, question answering and keyword extraction — but they run entirely on your own device using transformers.js and WebAssembly. Unlike a cloud AI service, the text you paste is never sent to a server or an API, and there is no account, key, or per-request charge.
The first time you open a tool it downloads a one-time model (from tens to a couple of hundred megabytes) directly from Whiztools, then caches it. After that, inference happens locally — you can even disconnect from the internet and it still works.
What these models are good at
On-device models are compact, so they are fast and private but not as capable as a large cloud LLM. They shine at focused, well-defined tasks: condensing an article, gauging whether feedback is positive or negative, pulling the people and places out of a document, answering a factual question from a passage, or surfacing the phrases that best represent a text.
- Summarizer — turn a long article or report into a short summary
- Sentiment analyzer — score reviews or messages as positive or negative
- Entity extractor — find people, organizations and locations
- Question answering — pull an answer straight from a passage
- Keyword extractor — rank the phrases that define a document
Where the limits are
Because the models are small enough to run in a browser tab, their quality is a step below GPT-class cloud models, and they will not write long original essays or translate with perfect fluency. Treat the output as a fast, private first pass rather than a finished product. For anything mission-critical, review the result — the trade you are making is total privacy and zero cost in exchange for using a lighter-weight model.
Browse related how-tos in our Guides, or read more about why Whiztools keeps processing on-device in About.