GitLab has released model 19.0 of its DevSecOps platform. The brand new launch focuses totally on additional integrating AI brokers all through all the software program improvement course of. With this, the corporate goals not solely to speed up code era but additionally to additional automate the guide steps concerned in evaluations, safety checks, and deployments.
Based on GitLab, a brand new drawback is rising inside many improvement groups: because of AI, builders are producing code quicker, whereas validation, testing, and deployment proceed to take a comparatively very long time. With model 19.0, the corporate is trying to scale back that delay by deploying AI brokers at a number of factors throughout the software program chain, studies TNW.
Central to that is the GitLab Duo Agent Platform, which turned usually obtainable earlier this yr. Within the new model, the brokers can carry out duties throughout varied phases of software program improvement, from planning and code era to safety remediation. This implies processes ought to run in parallel extra typically fairly than by means of sequential guide hand-offs between groups.
GitLab Duo Developer, the AI assistant for builders, can be gaining new capabilities. Customers can now immediately hyperlink the assistant to points or merge requests. Moreover, the agent may be invoked inside dialogue threads, after which it may possibly independently generate modifications or proposals.
SBOM scanning goals to enhance safety
Some of the vital additions is the final availability of dependency scanning primarily based on SBOMs, or Software program Payments of Supplies. This permits organizations to detect vulnerabilities inside complete dependency buildings, together with transitive dependencies that aren’t explicitly included in initiatives.
The performance helps Maven, Gradle, and Python environments, amongst others. Based on GitLab, third-party code poses an more and more vital safety danger in trendy software program improvement, partly as a result of improvement groups rely closely on open-source parts.
GitLab can be additional increasing help for AI fashions. Model 19.0 now helps, amongst others, Anthropic’s Claude Opus 4.7 and Google’s Gemini fashions. Open-source alternate options comparable to Devstral 2 and GLM-5.1 can be deployed inside self-hosted environments.
The Gemini integration is used for code evaluations, restoring CI/CD pipelines, and resolving safety points, amongst different issues. GitLab additionally helps self-hosted implementations of Mistral AI fashions.
For bigger organizations, GitLab is introducing organization-wide overview insurance policies. This implies settings now not should be configured individually for every repository. This could cut back administrative overhead, significantly for organizations with numerous initiatives.
The discharge additionally consists of technical modifications which will impression self-managed installations. For instance, Valkey replaces Redis because the default in Linux packages, the included Mattermost set up is eliminated, and help for Ubuntu 20.04 ends.