Montevive Labs: AI demos that run in your browser
Code you can try
Code you can try
Three open demos, no sign-up and no backend. Nothing you type leaves your machine: the model downloads into your browser and the processing happens there. It is the fastest way to check that what we say about local processing is not a figure of speech.
- 01Shieldstral Policy Probe. Write a moderation policy as a yes-or-no question and watch Mistral's open safety classifier answer in a single token. Then the same sentence in five languages: it claims Spanish support, but not Catalan, Basque or Galician.
- 02OpenAI Privacy Filter. Personal-information detection running entirely in the browser — names, emails, phone numbers, addresses and secrets — with the exact span of each detection and its confidence.
- 03ALIA tokeniser comparator. Side by side, how the ALIA, Llama 3 and Mistral tokenisers split the same text. It shows ALIA's efficiency in Iberian languages, which translates directly into cost per query.
We publish them out of conviction and out of method: working in the open is how a claim gets verified, and the same team that audits AI security needs to build it. Every demo came out of a real project.
Discover Labs at labs.montevive.aiWe build the technology before we sell it
Shieldstral Policy ProbeWatch live how Mistral's safety classifier decides which prompts to block and why, across languages.Open demo →OpenAI Privacy FilterDetect and anonymize personal data before it reaches an external LLM.Open demo →ALIA Tokenizer ComparisonCompare how ALIA and other reference models tokenize Spanish.Open demo →
Let's talk
The risk you don't see, and the opportunity you're not seizing yet
We reply in less than 24 hours.
ISO 27001GDPRNIS2Reply in <24h
Request your diagnostic
We'll reply within 24h.