About
I've worked with MobilSense since 2002 and am its CTO. In 2026 I directed the migration of MobilSentry, its flagship product, from leased data-center hardware to AWS, onto a platform managed through GitOps against an infrastructure-as-code control plane: five months from first commit to cutover, as a team of one directing AI agents. By cutover the project had 4,825 commits and 1,581 managed AWS resources. Customers saw just over six hours of outage against a twelve-hour window, and platform spend fell by about 70%. Every Role but Author tells how.
Before that, I led a 110-person development organization at CSG Systems building a convergent billing system, led development and system operations at Big Planet as vice president, and served as fractional CTO for AmeriVon and as partner and CTO at InstaSong.
Why Cycle Time
When we founded Terrapin Technologies in 1997, development time was driving schedule and budget overruns, and a lack of discipline and effective methods was a large part of the reason. We helped clients shorten their development cycle time with defined, repeatable processes. Cycle time is the elapsed time from starting work on a change to delivering it to the people who use it. A shorter cycle only matters if each one moves the system from one stable, known state to the next and converges to done. For traditional software development, the industry has since produced more than enough methodologies to go around. AI agents have reopened the question of how: how to deliver systems that meet their requirements, on time and on budget, to the quality expected, when agents write much of the code. Agentic development has been adopted faster than its methods have matured. cycletime.com is where I write about closing that gap, from the record of doing it.
Speaking
I speak on directing AI agents on production systems, keeping humans in charge of the gates, and modernizing long-lived platforms.