Your agent shouldn't have to wait for CI to find out something broke. 👀 Here’s how Chunk sidecars brings testing into the development loop, so Claude Code can catch and fix issues earlier 👇

More code in progress doesn't necessarily mean more code shipped. Our latest Q3 data shows why managing unfinished work is becoming a challenge for AI-assisted development, and what teams can do about it: circle.ci/4jUfrtp

More branches in flight doesn't always mean more changes landing on main. Our Q3 State of Software Delivery Pulse Report is out. Dig into the latest findings: circle.ci/4iAs4aL

Your agent can run relevant tests without pushing a single commit. Chunk Sidecars + Smarter Testing bring targeted validation into your agent's workflow, catching issues before they ever reach CI.

What are engineering leaders seeing as software delivery changes? That’s what @z00b digs into on our podcast, with convos about what’s changing, what teams are trying, and what they’re learning along the way. Listen on the Confident Commit Substack 👇 confidentcommit.com/

There’s understanding the problem, and then there’s getting your hands on it. As more people get the chance to build, that connection to the work gets a whole lot more real: circle.ci/4gzETRv

🗽 NYC! We're teaming up with @QodoAI to talk about what happens when AI agents move beyond the demo and into production. Join us on October 22: luma.com/l7dhbis5

Smarter Testing is officially GA! 🎉 Because running all 9,716 tests when only 44 actually matter? We can do better than that. In this demo: 3:56 minutes → 22 seconds Learn more and get started: circle.ci/4d1L1B1

Build the product → build the machine to sell it. That used to work, but when the tools can change in 48 hours, you have to keep building and learning as you go: circle.ci/4gzETRv

Turns out telling coding agents what to do over and over again doesn't actually make them better at fixing CI failures. Here's what worked instead (and cut token usage by 25%): circle.ci/474f2Nh

Loop Lab is where CircleCI engineers share the experiments happening inside their own work. What worked, what didn’t, and what they learned along the way, with ideas you can take back to your own workflow. Follow along on the Confident Commit Substack ⬇️ circle.ci/4An1EBe

Happy SF @Techweek_! We've got plans for you tomorrow. 🎉 Join our CTO @z00b for a hands-on workshop on getting AI-generated code to production with confidence. Lunch + coffee from Cento, a keyboard raffle, and field eng office hours included. RSVP: circle.ci/4A4Vo0D

Plot twist: the CEO is writing specs and the CTO is writing code again. When everything is changing this fast, sometimes you’ve gotta get back in there and build: circle.ci/4gzETRv

Your existing dev stack isn’t going anywhere, so where do AI tools fit? Tune into the full webinar: circle.ci/3UjSRzq

SF Tech Week is almost here! 👏 Angeline has the rundown on what we’ve got planned on October 6th, including a hands-on workshop, lunch + coffee, and a keyboard raffle. Come hang with us: circle.ci/4A4Vo0D

Before you add another AI tool, draw out your current stack. ✏️ Find the constraints first, then decide where AI can make a difference: circle.ci/3UjSRzq

We might not fully realize what kind of moment we’re in until later. Judson Griffin on why now’s the time to learn fast, get better at what you do, and embrace a little discomfort along the way: circle.ci/4Aao4FS

An agent wrote a script to archive hundreds of GitHub repos. It looked solid, except the GitHub CLI command it used didn’t exist. 😅 Olaf joined Beyond Vibe Coding to talk about what happens when code generation outpaces validation: circle.ci/4gWEJFe