Rob Zuber is a 20-year veteran of software startups; a three-time founder, and five-time CTO. Since joining CircleCI, Rob has seen the company through multiple rounds of funding and an acquisition, while leading a team of 200+ engineers distributed worldwide.  While he’s not solving complex software problems, Rob enjoys snowboarding, playing the guitar, and spending time with his wife and two children. He holds a Bachelor’s degree in Applied Science from Queen’s University in Kingston, Ontario, and lives in Oakland, California.

1. AI Code Assistants: Throughput vs. Value

Across CircleCI’s customer base, Rob is seeing clear signs of increased throughput thanks to coding assistants. “We generate a lot of code very quickly… we do see some uptick in just throughput, as I would describe it.”

But he cautions that throughput isn’t the same as value: “I’m building more. I have more throughput, but am I delivering more customer value? I like to believe that we will get there, but only time can confirm it.”

2. Shifting Bottlenecks in the Pipeline

AI has made it faster than ever to create code, but that just shifts the bottleneck downstream. “As you eliminate a bottleneck, the next one is going to appear.

Continuing to optimize how we write code, if that’s not the bottleneck, doesn’t actually produce additional value to the organization.” Rob warns that the old linear sequence of writing, reviewing, and merging starts to break down when AI can generate massive diffs in hours, leaving review and integration as the new chokepoints.

3. Rethinking “Good Code” in the AI Era

Is “good code” still the right yardstick? Rob points out that software has been optimized for humans for 50 years: readability, maintainability, extensibility.

But now, “we’re asking a machine to do the work. So would we optimize it the same way? I doubt it.” He suggests that if code can be generated and replaced on demand, “do I care about readability and maintainability? Or do I just trust the machine to create the inputs and outputs that I desire?”

4. The Risks of Vibe Coding

“Vibe coding,” using AI-generated code without fully understanding it, is already here. Rob compares it to developers pasting code from Stack Overflow: “It came up as the first answer, it seems to work. Well, how does it work? I don’t know.”

The risk is when this code lands in production: “Now production is down and an SRE is getting paged… but they might not understand the code themselves.” In this world, experienced SREs who can debug under pressure become even more essential.

5. Future of Engineering Roles

Rob closes with advice for engineers navigating this transition. “So much of it hinges on identity and what we believe our role is as engineers.”

He encourages people to know what drives them: “I frame what I do as applying my understanding of technology to solve customer problems.” For those worried about staying relevant, his advice is simple: “Know who you are and stay curious.” The craft of writing code is changing fast, but the need for problem solvers isn’t going away

‍