Real systems, real results.
Not mockups or concepts. These are working systems, including one I used to run the response to a high-stakes legal case on my own home.
SigilOS: a multi-agent AI operating system
A personal AI operating system where multiple specialized agents work together through one simple web dashboard. It put real AI capability behind a normal interface, with safety switches and a full audit trail. When my home became a construction-defect case, SigilOS ran the response on real documents and real deadlines.
- Extracted and organized email across multiple accounts
- Converted legal documents into clean, searchable notes
- Built a timeline measured against statutory deadlines
- Mapped evidence to the legal theories that mattered
- Rendered the property in 3D from 2D maps
- Coordinated attorney prep with a complete activity log
Let it Resolve: a production platform
A complete platform serving a passionate hobbyist community, live in production. It shows the full arc of taking an idea to a polished, fast, reliable product that real people use every day, including its own AI helpers that stay within clear boundaries.
- Live and serving real users in production
- Fast, polished interface built for everyday use
- AI assistants with clearly scoped, safe roles
Governed AI development
A way of working where the rules an AI follows are written down, version-controlled, and visible. Instead of hoping the AI behaves, its boundaries and conventions are explicit and auditable. The result for a business is AI work that is consistent, reviewable, and tailored to how you actually operate.
- AI behavior that is consistent and predictable
- Rules that are visible and reviewable, not hidden
- Tailored to each business rather than one-size-fits-all
Want a system like this for your work?
Tell me what you are facing. We will figure out whether a system like these would help, and what it would take.