A simpler way to run AI agents in Bitbucket Pipelines

AI agents in Bitbucket Pipelines previously required extensive manual configuration for compute, permissions, and runtime settings, often leading to repetitive YAML entries across steps. Workspace admins can now enable built-in AI agents directly from Workspace settings under AI → Agentic Pipelines, with Bitbucket automatically supplying the necessary underlying configuration. This centralizes setup while allowing repositories to retain control over runtime settings, such as build images or step sizes, or to disable specific agents when needed.

Reusable step configurations further streamline the process by letting teams define a named set of Pipelines options—such as images, caches, or permissions—once and apply them across multiple steps. This eliminates redundant configuration while keeping individual steps explicit and reviewable, reducing maintenance overhead for both AI-powered and standard build, test, or release tasks.

The new approach ensures that when a workspace admin enables a built-in agent, Bitbucket provides the agent and its required setup, while repository owners can still customize how the agent runs in their specific repository. This balance between centralization and flexibility aims to make AI agent adoption more practical without adding unnecessary abstraction layers.

The changes are part of broader updates to Bitbucket Cloud, including draft pull requests and branch deletion controls, though the AI agent enhancements specifically target the complexity of configuring AI-powered workflows in Pipelines.