Seat Time: a production platform, with AI as the pit crew.
Seat Time is my own company, not a client. It’s lap timing and driver profiles for track days and club racing, and it’s where I prove out running AI as leverage before I bring it to client work.
FROM THE SEAT TIME CODEBASE AS OF OCTOBER 12, 2026. TEST COUNT FROM A FULL LOCAL RUN.
Lap timing that keeps running when the internet doesn’t.
At the track, a desktop app reads the timing hardware and works out lap times, and it keeps timing even when the internet drops. In the cloud, organizers set up events, schedules and rosters. Drivers, organizers and spectators follow timing live on their phones, and results, analytics and leaderboards stay on each driver’s profile.
Timing real track days since March 2026.
Hardware plus cloud
The timing app talks to four different timing-decoder protocols and syncs both ways with the cloud.
Offline-first
Laps are saved locally and uploaded when the connection returns, without losing or duplicating any.
Many organizers, one platform
Each organizer gets its own public page, staff roles and permissions.
Safe releases
New versions go out to a canary first and switch back automatically if something fails.
Still in progress: online payments and the full registration system.
How I built it
I authored more than 95% of the 2,000+ commits in Seat Time’s codebase (as of October 12, 2026). Not by typing faster: by running AI the way a crew chief runs a pit crew. I set the plan, hand out the work, review what comes back, and decide what ships. The AI tools are leverage. The judgment stays with me.
One other developer contributed too.
Written rules
Every AI session follows the same working rules, including what needs my approval before it touches production.
Repeatable jobs as skills
Jobs like production deploys are written up as skills, so the agent follows the same checklist every time.
Parallel sessions
Several AI sessions run at once, coordinated so they don’t collide, like a crew working several jobs on the car.
AI checking AI
Staged whole-codebase reviews where some agents flag problems and others challenge the findings, before I decide what to act on.
This is how I work, not a guarantee on its own. AI reviews help surface problems; the call on what’s right, and the responsibility for it, stay with me.
WHAT IT MEANS FOR YOU Want this kind of building on your team?
That’s the leverage I bring to your team: AI doing the heavy lifting, with me setting the plan, reviewing the work, and taking responsibility for delivery. We define what each automation is allowed to do, test it before it goes live, and agree upfront which actions require human approval.
Request a free consult