We just “hired” a new AI Developer - Devin

No, not a human working on AI, but an AI Teammate. It’s been a fascinating experience giving a taste of what’s to come.

At Dagger, we’re a small team building a complex engine with an ever-growing list of use cases. Like many open-source projects, we face the challenge of maintaining the “long tail” of issues—all those small but important tasks that often pile up because they’re not critical enough to prioritize. That’s where my head went when I heard about Devin, the AI Teammate.

A Typical Open-Source Problem

Here’s a familiar story: Someone reports a somewhat minor inconvenience—maybe something annoying but probably not urgent. We care a lot about polish at Dagger but the list of things to do is long. It gets logged, but it’s not a priority, so it languishes. Three months go by, and no one’s had time to even look at it. At best, it gets a quick triage. At worst, it disappears into the abyss of GitHub issues.

Case in point: Issue #8195. A contributor highlighted a small but valid pain point in our workflow. Without Devin, it would’ve stayed in backlog limbo. But with Devin now on the team, we asked, “Can you take this?” And it did.

A New Kind of Contribution

Within minutes, Devin opened a pull request. The PR required some human review, but the core implementation was functionally correct on the first try. Devin even followed up on our feedback, iterating until the PR was ready to merge.

It felt like a solid contribution from a smart junior dev—a developer new to the codebase but eager to learn and improve. Except Devin is an AI. And it was working in a way that felt completely natural.

If you’re curious, you can watch the full session here. You’ll see how it worked through the issue, adapted to feedback, and produced a meaningful result.

Teaching Devin to Develop Dagger

Devin is different from your typical developer but managing Devin is in some ways familiar. Devin is very “book smart” but not yet “street smart” just like some super smart junior developers. One key to getting value is being smart about both the tasks you give Devin but also the way you train Devin. One of the most fascinating things about Devin is how seamlessly it learns.

We provided feedback both in the app and directly on GitHub, and Devin handled both methods without missing a beat.

Then we took things further: We taught Devin how to use Dagger to develop Dagger. Because our build and test environments are fully containerized with Dagger, Devin doesn’t even need CI. It can run its own CI locally, validate its work, and report back in a comment. When humans need to reproduce results, we just use the same containerized environment Devin already configured!

This wasn’t just a time-saver—it was a holy sh*t moment. It completely reframes how we think about development workflows. Devin’s ability to autonomously test, validate, and iterate locally changes the way we collaborate, and it hints at what the future of DevOps could look like.

A New Paradigm for Open Source

If you’re running an open-source project with any amount of maintenance load, you can’t afford not to take a look at Devin. It’s not just a tool—it’s a new kind of teammate, one that’s always ready to take on the long tail of tasks you never have time for.

Devin doesn’t replace developers. It amplifies what we can do, taking care of the repetitive and mundane so we can focus on what matters most. It’s a productivity boost, a fresh perspective, and a way to keep projects moving forward when bandwidth is tight.

What This Means for DevOps

Devin’s ability to autonomously contribute, run local CI, and adapt to human feedback isn’t just a cool feature—it’s a glimpse into the future of software development. With the right guardrails Devin can add a lot of value today, but long term, this technology may fundamentally change the relationship between developers and their tools, enabling us to build and ship software in ways that were previously unimaginable.

Thanks to the Cognition team for engaging with us on this: Congratulations on launching such a useful and groundbreaking product. Devin is already shaking up how we work, and I have no doubt it’s going to leave a lasting mark on the DevOps industry.

If you’re curious, give Devin a try. Your open-source backlog will thank you.