I come from operations and security, and I build solutions on language models myself. So I start with the questions that decide whether an AI solution holds up in operation: Where is the data? Who has access? What happens when the model gets it wrong?
What I can help with
- A realistic picture of where AI adds value for you, and where it does not.
- AI on local language models, so no data leaves your infrastructure.
- Security, access management and GDPR around the AI solutions.
- Governance for AI agents in development and operations: one owner per task, everything documented and independent checks before anything is called done.
How I build
I designed everything below, and it is built together with a standing team of AI agents that I lead. They work to the same rules I use in IT operations: every task has one owner, everything is documented, and an independent agent checks the work before it is called done. Data that must not leave the building is handled by language models on my own hardware. That is the experience I bring when a business starts using AI.
What I have designed and built
- A daily news briefing on local language models: It reads 84 news sources from around the world, groups the coverage of the same event across outlets and languages, compares what the outlets say and checks the claims before it is sent.
- A security watch (IDS): It checks servers, networks and logins every ten minutes and raises the alarm when something new or suspicious appears.
- Email monitoring: Every email the systems send gets a spam score and the DMARC reports are read daily, so misuse of the domains is caught.
- A private notification and login service: It brings alerts and reports from every system to phone and watch, with sign-in by one-time codes. It is also the sign-in for the client portal on this site.
- A professional tool for perfume formulation: It calculates IFRA and EU limits, allergens and hazard labelling, reads the suppliers’ safety data sheets and produces the documentation for the safety assessment.
- An app for motorcycle groups in the App Store and on Google Play: group intercom, shared navigation, a live map of every rider, road closures across Europe, crash detection and SOS, and route planning via a language model.
- Voice control of a large music archive: You say what you want to hear, and a language model finds it in the library and plays it in the right room.
- My own phone apps: among them a mail client and an app that gives every app its own volume.
Training
N2IT AI Academy
One complete training programme that takes a company from the first question to AI as a regular part of everyday work. Eight courses in four tracks: understand, work, govern and adopt. Every module has a presentation, an exercise with real files and a check that shows it works.
8 courses in 4 tracks · 48 modules · about 36 hours: Understand, Work, Govern, Adopt.
N2IT runs and facilitates the courses, on site at your premises or online, and can take charge of the whole introduction of AI in the company, from mapping to operations.
Tell me what you are facing
A few lines on the role, the timing and where you are based are enough. I reply personally.