Most engineers dream of building something transformative. Kit Colbert actually did it. Twice.
As an IC developer at VMware in the mid-2000s, he worked on vMotion, the technology that lets you move running virtual machines between physical hosts without anyone noticing. Later, he largely built Storage vMotion himself, a feature that turned multi-month, multi-million-dollar storage migrations into automated drag-and-drop operations.
But the technical achievement wasn't what mattered most.
In his conversation with Lucas Hendrich on CTO2CTO, Kit, now Platform CTO at Invisible Technologies, shares a crucial lesson from that experience: invention and innovation are fundamentally different things.
That distinction, learned 15+ years ago in virtualization, now shapes how he thinks about AI's impact on software teams, organizational structure, and what it means to create real value in 2026.
Kit describes himself as "a little renegade" early in his career. When he built Storage vMotion, he tried to circumvent VMware's ECR (Engineering Change Request) process—the formal approval mechanism for adding unplanned features to a release.
He lined up QA resources through personal favors, planning to ship the feature "under the table." But one critical person couldn't help, forcing him through the official route.
"I spent probably three or four months just trying to find someone who got this idea," Kit recalls. "Eventually I found a product manager who got it. His eyes lit up. He went to a VP who could make the call."
Once Storage vMotion entered the release formally, everything changed.
Marketing came to understand positioning. Tech support built runbooks. Sales learned how to pitch it. Pricing teams figured out packaging. The entire company—thousands of people, millions in infrastructure—mobilized to turn this invention into customer value.
"I would've had to go out there and pitch it to customers myself," Kit explains. "I would've had to support it. Instead, my personal effort went way down once I educated everyone. They took the ball and ran with it."
The lesson? Invention is about creating something new. Innovation is about impact.
A software organization is a machine designed to create value for customers. That requires marketing, sales, support, operations—functions that seem like "red tape" to individual engineers but are actually the mechanisms that transform code into business outcomes.
"If I'd gotten that last QA person to sign off and shipped it without the process," Kit reflects, "no one would've known about it. Maybe a customer would've found it and screwed something up. Who are they gonna call? Tech support doesn't know about it. Maybe they track me down at 2 AM."
The ECR process wasn't holding him back. It was preventing wasted effort downstream and ensuring the company could actually deliver value at scale.
Fast-forward to 2026, and Kit is watching a similar transformation play out—this time driven by AI.
At Invisible Technologies, a ~500-person startup, Kit sees developers using AI agents in ways that would've been impossible even a year ago. The implications for team structure are profound.
"The notion of the two-pizza team is gonna go away," he predicts. "You're gonna get maybe a half-pizza team: two or three people. When you have 20 or 30 agents in aggregate between these people, how do you manage all that code being created?"
The math is straightforward. If a developer can run 5-8 agents in parallel, each working semi-autonomously, the traditional 6-8 person team becomes obsolete. Smaller teams with agent leverage can produce dramatically more output.
But this creates new pressures:
We're already talking to our CFO about this," Kit says. "By the end of 2026, we expect to assume a developer is twice as expensive as their salary—because they're spending a salary's worth of tokens per year."
Rather than limiting token consumption, the focus is on smart usage. If $6,000/month in AI tools makes someone 10x more productive, that's a bargain compared to hiring additional engineers.
One of Kit's boldest predictions: by the end of 2026, AI agents will be able to run autonomously for a week.
"Today, the rating is around 12 hours," he explains. "You can set an agent going in the morning or overnight and check on it 12 hours later. My sense is that by year-end, you could give it a complex project plan and check back next Monday."
This isn't science fiction. It's an extrapolation based on current capability growth and new model releases.
"Think about how we manage humans," Kit continues. "A lot of my team, I only meet with them weekly because I know they're capable. We check in on progress, deal with issues, work through problems. You can imagine AI agents being the same: you're checking in on them weekly, validating their work."
The shift from managing people to managing agents (or managing people who manage agents) will require new leadership skills. Technical expertise matters less than the ability to set clear direction, prioritize effectively, and validate outputs intelligently.
A recurring theme in Kit's thinking: most of the time, we're wrong about what will create impact.
He cites Microsoft Research findings showing that two-thirds of software changes meant to improve a KPI either make no measurable difference or make things worse.
"If you believe you're going to be wrong a lot, it doesn't mean you shouldn't do things," Kit says. "But it means you should limit the effort you put into each thing and close the loop as quickly as possible."
This insight, central to Agile development, becomes even more important when AI can generate code at unprecedented speed.
Speed without validation is just noise. The ability to experiment rapidly, measure impact, and kill bad ideas before they consume resources will separate high-performing teams from those drowning in AI-generated output.
Kit's path to CTO wasn't a smooth climb. After a decade as an IC developer and tech lead, he moved into a CTO role on a smaller team, then took a GM position leading VMware's cloud-native apps business unit during the Docker/container disruption.
"I totally messed it up," he says candidly. "I spent two years slogging through that. It was awful and very stressful."
But he calls it the most educational experience of his career. "I was used to being an IC, where you lead through influence. I was good at setting vision, but I struggled with being firm and making decisions. I wasn't used to that. It took me a while to realize, oh shit, I gotta make the decision."
After that role, he went back to being a CTO with a smaller team, then eventually became CTO for all of VMware, managing 2,400 engineers. "I didn't have this progression of going from slightly bigger team to slightly bigger team," Kit explains. "It went from 150 people to 15 to 2,400. That weird leap was possible because of what I learned from failing."
The lesson? "I don't see failure as a bad thing. What's more important is: are you learning and getting better?"
As a leader at scale, he had to let go of technical decisions and focus on organizational, strategic, and priority calls that had downstream technical impact.
"Your value to the organization is different than it was before," he reflects. "What got you here won't make you successful going forward."
When asked what's overrated in the AI landscape, Kit doesn't hesitate: the idea that you can just throw AI at people and expect productivity gains. "AI is a normal technology. It's like any other tool. When you give a developer a new tool, they'll be less productive initially. It takes time to figure out how to use it, incorporate it into workflow, and get value. Just like Agile: you can't change tomorrow. You need structure and best practices."
He also pushes back on AGI hype: "This notion that some AGI can be better than the aggregate of all humans in every possible way doesn't make sense. Chimps are better at game theory than humans. We're not smarter than every other animal species in every way. AGI will be the same: really great in some ways, probably not in others."
What's underrated? How much AI is already changing organizational structure at startups. "The future is here, but it's not evenly distributed," Kit says. "At Invisible and other startups, we're experimenting constantly. We're rethinking how we structure teams. That's happening now, not theoretically. It'll come to larger companies, but it takes time. We have to prove it out first.
He also believes the ability of AI agents to operate autonomously for extended periods is massively underrated. "Today it's 12 hours. By year-end, it could be a week. That's extraordinarily underrated in terms of how much it'll change how we operate."
Kit Colbert's career offers a unique vantage point on technology adoption: from low-level systems work on vMotion, to cloud-native disruption, to now building AI-first products.
The through-line? Innovation isn't about technical brilliance alone. It's about impact. And impact requires understanding the human and organizational systems that turn inventions into value.
As AI transforms software development, the leaders who succeed won't just be those who adopt the latest tools fastest. They'll be the ones who rethink team structure, prioritize ruthlessly, validate relentlessly, and build organizations designed to amplify human creativity—not replace it.
Listen to the full episode of CTO2CTO to hear Kit's insights on VMware's evolution, AI's organizational impact, and why the movie Her still holds up as the best representation of human-computer interaction.
Let's talk about where you are today and where you want to go - our experts are ready to help you move forward.
