Why Qastel & why now
AI adoption has outrun the controls around it. Why we're building Qastel to give companies a safer way forward.
I'm Nordine, co-founder and CTO of Qastel AI. I spent four years shipping AI features at Microsoft. I left to build Qastel because companies have adopted AI faster than security and compliance teams can govern it.
AI is already inside the company
Open a browser inside a company and you will find ChatGPT tabs, office copilots, agents connected to the CRM and a long tail of tools that no one reviewed. Most use starts with a real business need. The risk arrives when customer records, contracts or source code cross into a tool without the right controls.
Why Qastel
Security teams need a control layer that follows data into the AI tools people use. Qastel discovers those tools, lets business teams describe rules in plain language, and enforces the rules at runtime.
A customer list can be redacted before it leaves the company. An unapproved model can be blocked. Each decision creates evidence for the security team and the audit trail.
Why now
AI use does not wait for a procurement cycle. A new app can spread through a team in an afternoon, while a policy review takes weeks. Agents raise the risk because they can reach more data and take action across connected systems.
Boards and regulators now ask companies to show which AI systems they use, what data those systems can access and who approved them. A PDF policy cannot answer those questions. Companies need controls at the point of use, before these workflows become infrastructure.
Building in public
I'll write about building Qastel from Paris and what we learn from security, privacy and compliance teams in the field. I'll also share how we build software with agents, including the work that still needs a human.
If you're a CISO, a DPO or a compliance lead wrestling with this, I'd like to talk: nordinemarie@qastel.ai.