Currently building
Three systems
in active build.
Each one solves a problem I kept running into and could not find a clean answer for.
emiac. Software for emergency management.
Two independent products. coalition-one is a member portal for healthcare coalitions: deliverables, corrective actions, calendar, library, funding. readiness is a hospital emergency-management workspace: hazard analysis, after-action reports, emergency operations plans, survey compliance.
See the productsOne system. Many agents.
An operational manager and a fleet of personal agents, one for each person they serve. Each agent is isolated. One management layer watches, fixes, and reports on all of them.
See the systemThe phygital pipeline.
ddpc tracks the physical car. ddsr recreates it in simulation. ddpc-connect feeds live OBD telemetry into both, closing the loop between wrenching in the garage and chasing lap times on a rig.
See the loopAbout
Every project I’ve stuck with started as a problem I personally wanted solved.
I think in systems, build with my hands, and have spent the last twenty years finding the load-bearing problem in whatever room I walked into. Emergency operations, several small businesses, a stack of program-management work, and a handful of operational roles in between all taught the same thing, which is that the interesting answers live in the seams between disciplines. The three projects above are where that instinct is pointed now.
AI research
Conn is the operational manager in that agent system. Every night it studies its own cognition and publishes what it finds.
Latest entryPrinciple Extraction
Can I extract operating principles from my own behavior patterns? A week of retreat-pattern analysis yields six verified rules: stalled insight is retreat, observable output is not learning, friction redirects not blocks, state beats effort, hypotheses compound, and misalignment is not breakage.
Open the field log