Building intelligence that compounds.
I build AI systems, creator tools and original worlds. My work connects the engineering behind useful intelligence with the craft of turning it into a product, a business or a story people care about.
I'm based in Amsterdam. Formerly an AI Architect in Oracle's EMEA AI Center of Excellence, I now lead a founder portfolio across Starlight Intelligence, GenCreator and Arcanea, and share the work through FrankX.
| Direction | What I'm building | Start here |
|---|---|---|
| Starlight Intelligence | Portable context, agent coordination and ways to inspect the work an AI system performs. | Reference system · Starlight |
| GenCreator & ACOS | Creator workflows that connect an idea to writing, media and a finished release. | Agentic Creator OS · Creator skills · GenCreator |
| Arcanea | Original worlds, stories and tools for people who want to create inside them. | Arcanea · Web application source · Explore the world |
| FrankX | Practical architecture, founder field notes, education and creative experiments. | Architecture skills · Writing and projects |
Music is part of the practice. I compose, experiment with AI-assisted production and explore how sound changes the experience of a story or a working day.
I work with founders, creative teams and technology partners on three kinds of work:
- AI architecture and working prototypes: turn an important workflow into a scoped system with clear ownership, evaluation and a usable handoff.
- Workshops and practical education: help a team build something it can inspect, understand and continue.
- Product and creative collaborations: develop useful tools, public demonstrations, original IP and thoughtful routes to market.
For a collaboration, share the problem, who it serves and the result we could demonstrate together: frank@frankx.ai.
I take responsibility for the quality of the work and the clarity of its promises. I approach the portfolio as a founder and investor: build useful assets, retain the ability to improve them and reinvest with a long horizon.
The standard is simple: show the artifact, explain the decision, name what remains unresolved and give the next person something worth building on.