Mark McCorkle
Engineering leader. Seventeen years building software organizations — now leading them through the shift to AI-native development.
I lead engineering organizations through the shift to AI-native software development. I've run that transition at real scale — a two-hundred-engineer division, every engineer onboarded onto agentic tooling, more than 80% still using it in any given week — and the most useful thing I can tell you is that the tools were never the hard part.
The hard part was context, operating models, and getting people to change how they work. Most organizations respond to AI with procurement: buy the tools, run a pilot, count the licenses. But a coding agent is only as good as the context you can put in front of it, and most codebases were written for readers who already knew where everything was. The organizations that win treat context and documentation as engineering infrastructure rather than overhead — they build repositories an agent can be dropped into cold.
Underneath that is a larger shift. When implementation gets cheap, value moves upstream — to framing the problem, choosing what to build, and knowing what good looks like. The scarce skill stops being writing code and becomes asking the right question before any code is written. Most of the failure I've seen in AI-assisted engineering wasn't bad output; it was a well-executed answer to the wrong question.
That changes what engineering leadership is. The job is no longer routing work to people — it's designing the system in which people and agents divide the work: deciding what gets verified deterministically and what still deserves human judgment, and building teams that get better at thinking, not just faster at typing. Leaders who scale control will lose to leaders who scale reasoning.
I test these ideas in the open — Nightgauge is where they run as code. If you're leading an engineering organization through the same shift, I'm easy to reach: mark@markamccorkle.com.
The record
accesso, 2019–present. Engineering Manager at a public attractions- and leisure-technology company. Sixteen engineers across four teams — mobile, backend, and platform — the broadest leadership span in a 622-engineer organization, shipping products used by millions of guests at major resorts and theme parks.
The AI program. Chair of the division's AI committee — strategy, responsible-AI standards, and adoption practice for roughly two hundred engineers. Founded the agentic coding initiative: 100% engineering onboarding, 80%+ sustained weekly active use, and a measured 10–15% capacity lift. Lead time for the largest stories fell from a full sprint to 3–5 days; pre-merge defect escape fell 40%.
Edibu LLC, 2015–present. Founder. A small product studio: BowlSheet, scoring analytics for competitive bowling coaches, on the App Store since 2011 — and the home of Nightgauge.
Transaction Data Systems, 2013–2019. Mobile engineering manager. Built and led the team shipping HIPAA-compliant healthcare mobile applications.
Before that. Shipping iOS since the App Store's first year, 2008.
Nightgauge
My working lab: an autonomous software factory, Apache-2.0 open core. Agents work the backlog on your own machines, and a deterministic pipeline of staged quality gates is the go/no-go gauge on everything they produce — documentation-first, portable across coding agents, cost-aware model routing. It builds itself, and every idea on this page gets tested there in running code before I'll argue for it out loud.
Writing
- The tools were never the hard part — July 2026