Cortex co-founder and CTO Ganesh Datta sits down with Arjun Iyer, CEO and co-founder of Signadot. Arjun explains why validating code before it merges has become essential now that coding agents are producing far more changes than teams can review by hand.

Arjun and Ganesh chat about why a merge should be treated as a contract, how agents can pick and run only the tests that matter for each change, and why the PR is replacing the developer as the unit engineering organizations need to govern. They also discuss where humans still belong in an agentic SDLC, from deciding what to build to staying accountable for what ships.

Cortex co-founder and CTO Ganesh Datta sits down with Naveen Puvvula, who recently left a director role at Life360 to become a senior principal engineer at construction tech startup Planera, to talk about what it takes to align both humans and AI agents around outcomes instead of activity.

Naveen and Ganesh discuss why architecture and product judgment remain squarely human tasks even as agents get better at execution, how organizational bottlenecks rather than raw compute determine whether AI investment pays off, and why engineering specialization is likely to stratify the way manufacturing did with the rise of factory technicians. They also get into the trust gap AI still needs to close in regulated, high-stakes industries, and why unlimited token budgets are reviving the same tech debt trade-offs teams used to make with headcount.

Cortex co-founder and CTO Ganesh Datta sits down with Adam Berman, who recently stepped back from VP of Engineering into an individual contributor role at Semgrep, the application security company built on static analysis. Adam and Ganesh dig into how AI coding agents are reshaping both engineering leadership and application security.

Adam shares why he traded running Semgrep's engineering org for hands-on building again, and how AI has changed the math for leaders who want to stay close to the code. He and Ganesh talk through why static analysis is having a resurgence as a cheap, fast guardrail for agents, how security review is shifting from CI into the IDE and even into the design phase, and how agents are getting unnervingly good at chaining together minor vulnerabilities into critical exploits. They close on how much to trust agents with fixing their own security issues, and what it actually takes to go "full dark factory" safely.