CTOs at Scale (Global): How can you trust your agentic engineers?
ONLY for CTOs (and VPEs / Heads of Engineering) running orgs of ~100+ engineers.
Brought to you by Gather.dev with support from our partners at Mergify.
How to Verify Your Agentic Code
A handful of teams have built software factories where most of the code is written by agentic swarms, not humans. As CTOs at Scale, you’ve probably seen the velocity in pockets: one team hits 10x, but it doesn’t spread, and the question that doesn’t go away is verification. How do you know the output is actually good, at scale, across every engineer, every repo, every team?
In May we broke down the elements of a software factory: orchestration, context, verification, clean-up, and compounding. This time we are going one layer deeper into the hardest of those: verification.
What is good software, and how do we know whether our agentic swarms are generating it? Can we test for functional correctness, performance, security and maintainability consistently across engineers, repos and teams? And what about maintaining architectural elegance?
I’ll be sharing some of the patterns that are working from CTOs that have successfully scaled agentic development, drawn from the interviews I’m doing for the upcoming O’Reilly book on Scaling AI Adoption in Engineering.
We’ll also have a quick interview with the CTO of Mergify on what verification looks like at the merge gate when most PRs are agent-authored, and how the gates have to change when code volume goes up 5 to 10x. We'll also have the CTO of Svitla sharing how they're seeing their clients scaling verification for agentic code.
This is a working session, not a webinar. After a brief opening to frame the problem, we’ll spend most of the hour in small breakout groups of CTOs tackling the hard verification questions: How do you test functional correctness, performance, security and maintainability consistently when the code is agent-written? What do you gate at the merge boundary and what do you defer to production telemetry? How do you preserve architectural elegance when no human reads the code?
We’ll close by sharing learnings from each group. If you’re looking for a talk you can half-listen to on mute, this isn’t it. If you want to discuss the actual issues you’re facing with CTOs who are solving the same problems, join us.
CTOs at Scale is an invite only community for CTOs / VPEs running larger orgs. It is a Gather.dev community, curated by Peter Bell.