Building Software You Can Trust with Chainguard
Tech Talks DailyNovember 21, 2026
3795
27:4920.58 MB

Building Software You Can Trust with Chainguard

How much do you know about the code your business runs, and what could you prove about it if a customer asked tomorrow?

On Tech Talks Daily, I'm joined by Quincy Castro, CISO at Chainguard. He describes open source as an act of generosity: people publish code so others can build on it. The security question is how a company takes advantage of that work while checking what goes into its software and how those components are built and maintained.

We discuss what useful evidence looks like. Quincy talks about identifying the original source, checking components for signs of tampering, and publishing a software bill of materials and a cryptographic signature.

A bill of materials tells you what went into the software, and a signature can help you verify its origin and integrity when checked against a trusted signer. Neither tells you on its own whether the code is safe to run. Quincy makes a similar point about supplier audits. They can give buyers useful assurance, and he also wants to see how a supplier's controls work when someone asks to look behind the paperwork.

I ask Quincy where security leaders should start when the list of vulnerabilities is long and taking a service offline comes with real costs. His starting point is what the business needs to keep running after an incident. Once you know which service matters most, you can follow it back to the systems, dependencies, and people responsible for it. We also talk about the gap between what an audit samples and what an attacker might try, and how lessons from incidents can feed back into the way teams build and maintain software.

The interview closes with Chainguard's approach to AI in software development. Quincy explains why the company built Harness to help test applications with advanced AI models and Workspaces to give developers and agents isolated environments in which to collaborate.

What interested me was the question this poses for any team introducing a coding agent: what can it access, what can it change, and who checks the result before it goes live? I'd love to hear your experience. If a customer asked you to prove where a component came from and how it was built, what could your team show them today?

Please share your thoughts.