Community
We share knowledge, not just results
Blog
Articles on safe AI, the AI Act and Shadow AI: practical guides and what we learn in every project.
Kev: the open model that classifies text and tells you when it is not sure
Kev reads a text and answers a fixed set of questions with probabilities, not sentences. We analyse an evaluation on a consumer GPU: where it is right, where it hesitates, where it does not fit and what it costs against the commercial alternative.
Read more →ISO 42001 in Spain: what it's really for, and who shouldn't bother
What the standard actually certifies, how it differs from the AI Act, what the auditor will ask for, and when certifying is a sledgehammer to crack a nut.
Read more →Shieldstral runs in your browser now. Does it work in your languages?
We converted Mistral's open safety filter so it runs inside your browser, with nothing sent to any server. Then we measured it across five languages. The failure we found is the kind you would never notice.
Read more →Open Source
We share tools, agents and experiments with the community. Open collaboration makes our AI safer, and everyone else's too.
Penpot MCP server
🚀 Autocache - Intelligent Anthropic API Cache Proxy Automatically inject cache-control fields into Claude API requests to reduce costs by up to 90% and latency by up to 85%. Works as a transparent drop-in replacement for popular AI platforms like n8n, Flowise, Make.com, LangChain, and LlamaIndex—no code changes required
High-performance Go library for PII name detection. 10x faster than Python with embedded 727K names dataset. Production-ready out-of-the-box functionality for privacy compliance and data security.
Prompt Injection Demo with n8n
The risk you don't see, and the opportunity you're not seizing yet
We reply in less than 24 hours.
Request your diagnostic
We'll reply within 24h.