In this episode, Lucas Hendrich, CTO at Forte Group, sits down with Kit Colbert, Platform CTO at Invisible Technologies and former CTO at VMware, to explore the difference between building something new and creating real impact, and why that distinction matters more than ever in the age of AI.
"Invention is about something new. Innovation is about impact."
After spending nearly 20 years at VMware, including pioneering work on vMotion and Storage vMotion, Kit learned this lesson the hard way. He built Storage vMotion largely on his own, a technology that could turn multi-month, multi-million-dollar storage migrations into a simple drag-and-drop operation. But without marketing, sales enablement, and tech support aligned, the invention would've created zero innovation. The "red tape" he tried to avoid was actually the machinery that creates customer value.
"By the end of 2026, AI agents will be able to run for a week on their own."
Today's frontier is around 12 hours of autonomous operation. Kit predicts that within months, you'll be able to give an AI agent a complex project plan on Monday and check back in the following Monday, just like managing a capable human team member. This shift will fundamentally change how engineering teams are structured and how work gets orchestrated.
"The two-pizza team is going away. We're moving to half-pizza teams."
When developers have 5-8 AI agents working in parallel, the traditional team size becomes obsolete. Kit sees a future where 2-3 people manage 20-30 agents in aggregate, creating enormous pressure on product management and task orchestration. The challenge isn't writing code: it's managing the volume of code being generated and ensuring it creates real value.
"We're already assuming a developer will be twice as expensive by year-end because of token costs."
Kit's CFO is planning for engineers to consume a full salary's worth of tokens annually. Rather than limiting consumption, the focus is on smart usage. If spending $6,000/month on AI tools makes a developer 10x more productive, that's a bargain. The economics are shifting rapidly, and organizational planning needs to catch up.
"Two-thirds of software changes don't make any measurable difference or make things worse."
Drawing on Microsoft Research findings, Kit emphasizes that being wrong is normal in software development. The key is limiting effort per experiment and closing feedback loops quickly. This insight, central to the Agile revolution, becomes even more critical when AI can generate massive amounts of code. Speed without validation is just expensive noise.
"I learned more from failing as a GM than from any other role."
Kit's journey from IC developer to managing 2,400 engineers wasn't linear. He jumped from 150 people to 15, then to 2,400. The "failure" as a GM taught him lessons about decision-making, letting go of technical control, and understanding that his value to the organization had fundamentally changed. Leading through influence scales differently than technical expertise.
Stay tuned for more conversations on engineering leadership, AI transformation, and building products that matter on CTO2CTO.