I lead engineering teams, and for the past several years much of that work has been bringing AI into how those teams build software. At Adage Technologies I led the adoption in stages. We piloted GitHub Copilot and Microsoft Copilot first, then ran a separate pilot for Claude Code and worked with IT to roll it out to development staff. Along the way I contributed to the guidance engineers worked from, helped establish a group focused on AI governance, and coached people on how to evaluate what the tools produce. Engineers reported 10 to 20 percent efficiency gains and more time for planning and code hardening. I use the same tools every day in my current consulting work.
The tools only help when the people using them are good, so most of my attention has gone to people. Earlier at Adage I managed teams averaging 10 direct reports, including tech leads and software architects. After a restructure I led frontend engineering company wide, managing 8 engineers across North America and Latin America and coordinating 5 to 8 backend developers on active projects. At least 10 of my direct reports were promoted.
I came up through frontend engineering, and I still set technical direction in React and TypeScript, lead architecture and design reviews, and review code. More than 18 years of building web platforms is what lets me judge whether a team's work, or an AI tool's output, is good.