I build systems to understand where AI helps, where the work gets harder, and what it takes to make an idea useful. These projects range from personal tools to product prototypes.
Personal AI operating system
PatrickOS
Active build
A shared memory and coordination layer for AI tools, so project context and decisions can follow the work from one assistant to the next.
The next productivity jump is not one smarter chatbot. It is a reliable system for keeping AI work coordinated across real projects.
What exists
The build includes a shared memory vault, dashboard, tool connections, and review of proposed memory updates. The public build note explains the problem and approach.
Some are products, some are operating systems, and some are experiments. The common thread is practical: build enough to learn what reality thinks.
Interactive prototype
Bike shop service workflow
SpokeDesk
A connected service-counter and repair experience for bike shops: work orders, mechanic queues, estimate approval, rider updates, and checkout.
Local operators do not need abstract AI. They need tools that reduce interruptions and make the next customer interaction easier.
What exists
The prototype connects work orders, estimate approvals, messages, and mechanic checklists using shared sample data. Payments and retail inventory are simulated.
The workflow
1Check in the bike
2Approve and complete the repair
3Update the rider and simulate checkout
Private-beta preparation
Software vendor decisions
ClarityLens
An AI-assisted workspace for software buyers to compare vendors, weight criteria, separate claims from evidence, and prepare a defensible recommendation.
A vendor decision should make its assumptions and tradeoffs visible to the people who will live with the result.
What exists
The build includes weighted evaluation matrices, saved comparisons, and decision-pack exports. It is being prepared for a private beta.
The workflow
1Set decision criteria
2Compare claims and evidence
3Export a recommendation
MVP exploration
AI risk and readiness product
Agent Passport Studio
A way to inventory AI agents, describe what they can touch, and turn that into an executive-readable risk and governance view.
As agents move into business workflows, leaders need a practical record of what each agent does, where it has access, and what could go wrong.
What exists
The design combines agent profiles, consistent risk scoring, and executive reports. It remains an MVP exploration.
Internal portfolio
Governance operating surface
AI Governance Portfolio
A collection of governance patterns, release gates, and leadership artifacts for moving AI work from scattered pilots into managed operating practice.
The organizations that win with AI will not be the ones with the most experiments. They will be the ones that can safely scale what works.
What exists
An internal working collection of governance patterns and release criteria.
Prototype
Workflow redesign tool
Agentic Value Stream Designer
A concept for mapping work as it actually flows, then identifying where agentic AI changes the bottleneck, handoff, or decision point.
AI transformation breaks when it starts with tools instead of work. This starts with the value stream and asks where intelligence should enter.
What exists
A prototype for mapping process waste and turning it into a proposed AI-agent backlog.
Ongoing writing
Writing and field notes
The AI Catalyst
The writing arm of the work: essays about enterprise AI, agentic systems, leadership, governance, and the future of work.
Writing is how I sharpen the point of view, test the arguments, and make the practical lessons easier for other leaders to use.
What exists
Complete essays are available in the searchable archive, with recent writing linked to its original publication.
The writing shows how I think. The projects show how I test that thinking against real constraints: users, security, deployment, confusing workflows, thin margins, and the ordinary mess of getting useful systems into the world.