Twenty-nine new projects from the last few years — production products, internal systems, brand books, and concepts. Filter by lens.
Showing 29 of 29
Trip safety for whole organizations, in one production platform.
A governed AI delivery system, in closed beta. The abyss is the stage; amber is the decision.
A 40-year clothier gets a brand system.
Public risk feeds turned into trip-specific alerts.
Seven systems consolidated into one field-service product.
A map you can zoom from world to alert.
Brand decisions you can audit, reasoned against 766 principles.
One component library, many brands, light and dark.
Provenance you can see, before you trust the answer.
The public case for reviewing every trip.
Discovery with evidence, gates, and a human at the end.
Where agent work gets evidence, cost, and approval.
Seven repositories, 21 shared packages, one platform.
140+ specialist agents, reproducible and versioned.
Not a brand guide — a brand that behaves.
A credible identity for a new AI product, in days.
Creative direction, tested blind against a control.
Adapt a manuscript from a source you can defend.
Take the tedious out of tabletop.
The right skill context, resolved deterministically.
Two weeks of research and publishing, down to hours.
Source documents to claim drafts, with consensus checks.
Compliance made legible for people who are not specialists.
Story into action for a global relief nonprofit.
A manufacturing ERP designed in Figma and taken through a 52-page prototype.
Repeatable AI image styles, not one-off prompts.
Brand assets and a homepage system for a deep-tech company's public face.
Design a custom sign before you buy it.
Translation for the moments that are not transactions.
The right skill context, resolved deterministically.
A Model Context Protocol server that gives agents the right skill context every time: it finds relevant skills for a task, resolves the full upstream dependency chain, and loads their templates and checklists.
A registry of 140+ agents with hundreds of skills is only useful if the correct methodology reaches the agent at the moment of work — deterministically, with its dependencies, and auditably.
Four tools with topological dependency resolution, cycle detection, reference loading, and append-only audit logging, so agent output quality is repeatable and inspectable. Used by the Serras and Convergence pipeline.
I conceived, architected, and led AI-assisted development.