Cursor has introduced Rollouts, an AI agent that tracks code changes through deployment and watches for regressions once they reach production, according to a blog post published Wednesday by Rustam Lalkaka, formerly CEO of Firetiger and now at Cursor following the company's acquisition of his startup last month. The new tool addresses what Lalkaka describes as an imbalance in modern software development: while AI coding assistants have made writing code faster and cheaper, the work of safely deploying those changes and spotting problems after launch hasn't kept pace.
Rollouts begins its work when a developer opens a pull request. It reviews the proposed code, identifies which systems might be affected, and creates a monitoring plan that outlines what the change should accomplish, what risks it sees, which signals it will track, and any gaps in available instrumentation. Developers can review and adjust that plan before deployment. Once the code ships, Rollouts examines telemetry data including logs, metrics, and traces, checking them against the plan it generated earlier. Staging and production environments are evaluated separately, with each deployment receiving one of three outcomes: verified healthy, regression detected, or inconclusive. If Rollouts identifies a regression, it can flag the suspected change, notify the responsible developer, and either open a revert pull request for human review or pass the issue to a Cursor cloud agent to attempt a repair. The tool doesn't automatically merge fixes or roll back deployments, though it can halt a progressive rollout.
Lalkaka writes that the effort to create software changes has "dropped to near zero" thanks to agentic coding tools over the past two years, while "the cost and risk of deploying them has stayed largely the same." He frames the problem as a shift in bottlenecks: "Writing code is no longer the slow part. What hasn't sped up is everything after the PR goes up: making sure code is secure, watching the deploy, deciding whether a latency bump is real, figuring out which of eleven changes broke checkout." Rollouts is effectively a rebuild of Firetiger's Change Monitors product inside Cursor, constructed using a newly released framework called Bot Development Kit. The report notes that Rollouts can already catch problems isolated to a specific endpoint or region before they set off broader alerts, and can tell the difference between expected behavioral shifts and genuine regressions.
The launch comes as Cursor builds out Origin, its Git-compatible code hosting platform introduced in August as an alternative to GitHub designed for teams working heavily with AI agents. Rollouts can connect to either Origin or GitHub for source control, pull deployment events from continuous delivery systems, and draw signals from Datadog and similar telemetry providers. When a regression surfaces, it can hand the problem back to a Cursor cloud agent for investigation or repair. The report explains that Origin gives Cursor a native environment for more of this workflow loop: its cloud agents can already create branches, commit and push code, and open pull requests against Origin repositories, with Rollouts now adding visibility into what happens after deployment. Cursor also unveiled an upgraded Security Reviewer bot alongside Rollouts, which analyzes pull requests automatically for exploitable issues like injection flaws and broken authentication, providing a severity rating, attack path, and proposed fix. Average review time has dropped 21%, from 4.8 minutes to 3.8, while developer acceptance of its suggestions has climbed from roughly 45–50% to 60–70%. Both tools are available to customers on Cursor's Teams and Enterprise plans through its Automations tab. The tighter integration between coding agents, repositories, security checks, and production feedback could become Cursor's main competitive advantage as rivals including Zed with its Delta platform and Datadog with Bits Release pursue similar territory. The real test will be whether proximity alone can persuade developers to migrate from GitHub's entrenched ecosystem to a platform still filling out its feature set.