I build open source tools, companies and communities that help organisations understand, manage and improve complex systems: safely, transparently and for the long term. What I am building now: • RiskMandate.ai: the business risk layer for autonomous systems. • Newsroom.sgit.ai: Field notes on AI agents, security and the next generation of publishing, by Dinis Cruz • Sgit.ai: encrypted git for humans and AI agents, Apache-2.0. The foundation the other four are built on; sgraph.ai commercialises it as SG/Send and SG/Vaults. • VoiceDebrief.ai: voice recordings into transcripts and debriefs • TheCyberBoardroom.com: an AI-powered platform for the conversation between security teams and the board. How I build: Open source is the best business model for an AI startup, and it is the one all of mine run on. Technology is not the moat; what a company sells is trust, maintenance, quality and certification, and everything else should be open so the buyer never has to take that on faith. Whether an open-source project survives is a property of its copyright and trademark structure, not its licence file, and sovereignty is only possible when the code and the schemas are open. The argument, with its counter-cases, is at open-source.sgit.ai. The practice is that my companies ship open source, and their investor materials are public too. Where I come from: 20+ years across cyber security and software engineering. Former OWASP Board member; organised the OWASP Summits (Lisbon 2011, Woburn 2017) that became the Open Security Summit. Created the O2 Platform and the OSBot / MGraph-DB family of open-source tools, still in production across my companies today. I previously worked as a CISO and CTO for other companies alongside my own, and I build GenAI-powered teams where security and engineering are accelerators for the business rather than gates.
