There's a pattern I keep seeing across enterprise engineering teams right now.

The developer experience side of the house is fully bought in on AI. Copilots, code generation, AI-assisted code review. Developers are using these tools daily and the productivity gains are real. JetBrains' January 2026 AI Pulse data confirmed it: a large majority of developers now use AI tools in their daily work.

Then you ask about the CI/CD pipeline, the system that actually validates and ships all that AI-generated code, and the answer is crickets.

AI adoption inside CI/CD pipelines significantly lags behind development workflows according to JetBrains' 2026 research.

Everyone's using AI to write the code. Almost nobody's using it to run the pipeline that ships the code.

The JetBrains research puts it plainly: the main challenge is not technical integration. Teams are evaluating whether AI can reliably and predictably deliver value within a system that is responsible for validation.

That's the key word: validation. Your IDE is a low-stakes environment. You can throw away a bad suggestion from Copilot with no consequences. Your CI/CD pipeline is not that. It's the last line of defense before software ships to production. The trust bar is fundamentally higher.

In practice, AI adoption inside pipelines is shaped more by trust and measurable outcomes than by model capability. The models are capable enough. The organizational trust isn't there yet.

There's also a structural reason nobody talks about. In most engineering orgs, the pipeline is owned by a different team than the developers writing code. DevEx improvements happen fast because developers feel the pain directly and advocate for their own tools. Pipeline improvements happen slowly because the people who feel the pain, the platform or DevOps team, are operating with a different set of priorities and a different budget cycle.

AI got adopted fast in development workflows because developers had direct access and low-stakes experimentation space. CI/CD doesn't have that.

Here's what the data shows is landing in pipelines today.

Test intelligence is the clearest win. Testing is one of the most resource-intensive parts of CI/CD, and AI is beginning to show measurable impact, identifying which tests are most likely to catch a given change, surfacing flaky tests, and reducing the noise that erodes developer trust in the build.

Security remediation is gaining ground too. AI is being used to interpret findings from existing scanning tools, suggest fixes, and in some cases generate patches, all while still passing through standard CI/CD validation steps. It's AI augmenting the pipeline, not replacing its judgment.

AI agents are also starting to handle CI/CD configuration directly. AI agents can now set up build configurations and even full build chains, with the gap between describing a pipeline and actually having it running now very small. That's a meaningful shift. Configuration that used to take hours of back-and-forth is converging in minutes.

The teams that get ahead of this aren't waiting for the ecosystem to mature. They're running small, scoped experiments: one AI-assisted test selection integration, one automated triage workflow for failed builds. They're generating data, measuring impact, and building organizational trust incrementally.

Teams want more affordable solutions, proven reliability, and clearer security frameworks before committing to wider adoption. That's a reasonable position. The answer isn't to force adoption, it's to run the pilot that generates the proof.

Pick one thing. Measure it. Present the delta.

The AI adoption gap in CI/CD isn't a technology problem. It's a trust problem. And trust is built one verified outcome at a time.