[{"data":1,"prerenderedAt":810},["ShallowReactive",2],{"/en-us/blog/inside-look-how-gitlabs-test-platform-team-validates-ai-features":3,"navigation-en-us":45,"banner-en-us":445,"footer-en-us":455,"blog-post-authors-en-us-Mark Lapierre|Vincy Wilson":696,"blog-related-posts-en-us-inside-look-how-gitlabs-test-platform-team-validates-ai-features":722,"assessment-promotions-en-us":762,"next-steps-en-us":800},{"id":4,"title":5,"authorSlugs":6,"body":9,"categorySlug":10,"config":11,"content":15,"description":9,"extension":31,"isFeatured":13,"meta":32,"navigation":13,"path":33,"publishedDate":22,"seo":34,"stem":39,"tagSlugs":40,"__hash__":44},"blogPosts/en-us/blog/inside-look-how-gitlabs-test-platform-team-validates-ai-features.yml","Inside Look How Gitlabs Test Platform Team Validates Ai Features",[7,8],"mark-lapierre","vincy-wilson",null,"ai-ml",{"slug":12,"featured":13,"template":14},"inside-look-how-gitlabs-test-platform-team-validates-ai-features",true,"BlogPost",{"title":16,"description":17,"authors":18,"heroImage":21,"date":22,"body":23,"category":10,"tags":24},"Inside look: How GitLab's Test Platform team validates AI features","Learn how we continuously analyze AI feature performance, including testing latency worldwide, and get to know our new AI continuous analysis tool.",[19,20],"Mark Lapierre","Vincy Wilson","https://res.cloudinary.com/about-gitlab-com/image/upload/v1750099033/Blog/Hero%20Images/Blog/Hero%20Images/blog-image-template-1800x945%20%2811%29_78Dav6FR9EGjhebHWuBVan_1750099033422.png","2024-06-03","AI is increasingly becoming a centerpiece of software development - many companies are integrating it throughout their DevSecOps workflows to improve productivity and increase efficiency. Because of this now-critical role, AI features should be tested and analyzed on an ongoing basis. In this article, we take you behind the scenes to learn how [GitLab's Test Platform team](https://handbook.gitlab.com/handbook/engineering/infrastructure/test-platform/) does this for [GitLab Duo](https://about.gitlab.com/gitlab-duo/) features by conducting performance validation, functional readiness, and continuous analysis across GitLab versions. With this three-pronged approach, GitLab aims to ensure that GitLab Duo features are performing optimally for our customers.\n\n> Discover the future of AI-driven software development with our GitLab 17 virtual launch event. [Watch today!](https://about.gitlab.com/eighteen/)\n\n## AI and testing\n\nAI's non-deterministic nature, where the same input can produce different outputs, makes ensuring a great user experience a challenge. So, when we integrated AI deep into the GitLab DevSecOps Platform, we had to adapt to our best practices to address this challenge.\nThe [Test Platform team's mission ](https://handbook.gitlab.com/handbook/engineering/infrastructure/test-platform/) is to help enable the successful development and deployment of high-quality software applications with continuous analysis and efficiency to help ensure customer satisfaction. The key to achieving this is by delivering tools that help increase standardization, repeatability, and test consistency.\nApplying this to GitLab Duo, our AI suite of tools to power DevSecOps workflows, means being able to continuously analyze its performance and identify opportunities for improvement. Our goal is to gain clear, actionable insights that will help us to enhance GitLab Duo's capabilities and, as a result, better meet our customers' needs.\n## The need for continuous analysis of AI\n\nTo continuously assess GitLab Duo, we needed a mechanism for analyzing feature performance across releases. Therefore, we created an AI continuous analysis tool to automate the collection and analysis of data to achieve this.\n![diagram of how the AI continuous analysis tool works](https://res.cloudinary.com/about-gitlab-com/image/upload/v1750099041/Blog/Content%20Images/Blog/Content%20Images/image1_aHR0cHM6_1750099041503.png)\n\n\u003Ccenter>\u003Ci>How the AI continuous analysis tool works\u003C/i>\u003C/center>\n\n### Building the AI continuous analysis tool\n\nTo gain detailed, user-centric insights, we needed to gather data in the appropriate context – in this case, the integrated development environment (IDE), as it is where most of our users access GitLab Duo. We narrowed this down further by opting for the Visual Studio Code IDE, a popular choice within our community. Once the environment was chosen, we automated entering code prompts and recording the provided suggestions. The interactions with the IDE are handled by the [WebdriverIO VSCode service](https://github.com/webdriverio-community/wdio-vscode-service), and CI operations are handled through [GitLab CI/CD](https://docs.gitlab.com/ee/ci/). This automation significantly scaled up data collection and eliminated repetitive tasks for GitLab team members. To start, we have focused on measuring the performance of GitLab Duo Code Suggestions, but plan to expand to other GitLab AI features in the future.\n\n### Analyzing the data\n\nAt the core of our AI continuous analysis tool is a mechanism for collecting and analyzing code suggestions. This involves automatically entering code prompts, recording the suggestions provided, and logging timestamps of relevant events. We measure the time from when the tool provides an input until a suggestion is displayed in the UI. In addition, we record the logs created by the IDE, which report the time it took for each suggestion response to be received. With this data, we can compare the latency of suggestions in terms of how long it takes the backend AI service to send a response to the IDE, and how long it takes for the IDE to display the suggestion for the user. We then can compare latency and other metrics of GitLab Duo features across multiple releases. The GitLab platform has the ability to analyze [code quality](https://docs.gitlab.com/ee/ci/testing/code_quality.html) and [application security](https://docs.gitlab.com/ee/user/application_security/), so we leverage these capabilities to enable the AI continuous analysis tool to analyze the quality and security of the suggestions provided by GitLab Duo.\n\n### Improving AI-driven suggestions\n\nOnce the collected data is analyzed, the tool automatically generates a single report summarizing the results. The report includes key statistics (e.g., mean latency and/or latency at various percentiles), descriptions of notable differences or patterns, links to raw data, and CI/CD pipeline logs and artifacts. The tool also records a video of each prompt and suggestion, which allows us to review specific cases where differences are highlighted. This creates an opportunity for the UX researchers and development teams to take action on the insights gained, helping to improve the overall user experience and system performance.\n\nThe tool is at an early stage of development, but it's already helped us to improve the experience for GitLab Duo Code Suggestions users. Moving forward, we plan to expand our tool’s capabilities, incorporate more metrics and consume and provide input to our [Centralized Evaluation Framework](https://about.gitlab.com/direction/ai-powered/ai_framework/ai_evaluation/), which validates AI models, to enhance our continuous analysis further.\n\n## Performance validation\n\nAs AI has become integral to GitLab's offerings, optimizing the performance of AI-driven features is essential. Our performance tests aim to evaluate and monitor the performance of our GitLab components, which interact with AI service backends. While we can monitor the performance of these external services as part of our production environment's observability, we cannot control them. Thus, including third-party services in our performance testing would be expensive and yield limited benefits. Although third-party AI providers contribute to overall latency, the latency attributable to GitLab components is still important to check. We aim to detect changes that might lead to performance degradation by monitoring GitLab components.\n### Building AI performance validation test environment\n\nIn our AI test environments, the [AI Gateway](https://docs.gitlab.com/ee/architecture/blueprints/ai_gateway/#summary), which is a stand-alone service to give access to AI features to GitLab users, has been configured to return mocked responses, enabling us to test the performance of AI-powered features without interacting with third-party AI service providers. We conduct AI performance tests on [reference architecture environments of various sizes](https://docs.gitlab.com/ee/administration/reference_architectures/). Additionally, we evaluate new tests in their own isolated environment before they're added to the larger environments.\n\n### Testing multi-regional latency\n\nMulti-regional latency tests need to be run from various geolocations to validate that requests are being served from a suitable location close to the source of the request. We do this today with the use of the [GitLab Environment Toolkit](https://gitlab.com/gitlab-org/gitlab-environment-toolkit). The toolkit provisions an environment in the identified region to test (note: both the AI Gateway and the provisioned environment are in the same region), then uses the [GitLab Performance Tool](https://gitlab.com/gitlab-org/quality/performance) to run tests to measure time to first byte (TTFB). TTFB is our way of measuring time to the first part of the response being rendered, which contributes to the perceived latency that a customer experiences. To account for this measurement, our tests have a check to help ensure that the [response itself isn't empty](https://gitlab.com/gitlab-org/quality/performance/-/blob/cee8bef023e590e6ca75828e49f5c7c596581e06/k6/tests/experimental/api_v4_code_suggestions_generation_streaming.js#L70).\nOur tests are expanding further to continue to measure perceived latency from a customer’s perspective. We have captured a set of baseline response times that indicate how a specific set of regions performed when the test environment was in a known good state. These baselines allow us to compare subsequent environment updates and other regions to this known state to evaluate the impact of changes. These baseline measurements can be updated after major updates to ensure they stay relevant in the future.\nNote: As of this article's publication date, we have AI Gateway deployments across the U.S., Europe, and Asia. To learn more, visit our [handbook page](https://handbook.gitlab.com/handbook/engineering/development/data-science/ai-powered/ai-framework/#-aigw-region-deployments).\n\n## Functionality\n\nTo help continuously enable customers to confidently leverage AI reliably, we must continuously work to ensure our AI features function as expected.\n\n### Unit and integration tests\n\nFeatures that leverage AI models still require rigorous automated tests, which help engineers develop new features and changes confidently. However, since AI features can involve integrating with third-party AI providers, we must be careful to stub any external API calls to help ensure our tests are fast and reliable.\n\nFor a comprehensive look at testing at GitLab, look at our [testing standards and style guidelines](https://docs.gitlab.com/ee/development/testing_guide/).\n### End-to-end tests\nEnd-to-end testing is a strategy for checking whether the application works as expected across the entire software stack and architecture. We've implemented it in two ways for GitLab Duo testing: using real AI-generated responses and mock-generated AI responses.\n\n![validating features - image 2](https://res.cloudinary.com/about-gitlab-com/image/upload/v1750099041/Blog/Content%20Images/Blog/Content%20Images/image2_aHR0cHM6_1750099041504.png)\n\n\u003Ccenter>\u003Ci>End-to-end test workflow\u003C/i>\u003C/center>\n\n#### Using real AI-generated responses\n\nAlthough costly, end-to-end tests are important to help ensure the entire user experience functions as expected. Since AI models are non-deterministic, end-to-end test assertions for validating real AI-generated responses should be loose enough to help ensure the feature functions without relying on a response that may change. This might mean an assertion that checks for some response with no errors or for a response we are certain to receive.\n\nAI-driven functionality is not accessible only from within the GitLab application, so we must also consider user workflows for other applications that leverage these features. For example, to cover the use case of a developer requesting code suggestions in [IntelliJ IDEA](https://www.jetbrains.com/idea/) using the GitLab Duo plugin, we need to drive the IntelliJ application to simulate a user workflow. Similarly, to ensure that the GitLab Duo Chat experience is consistent in VS Code, we must drive the VS Code application and exercise the GitLab Workflow extension. Working to ensure these workflows are covered helps us maintain a consistently great developer experience across all GitLab products.\n#### Using mock AI-generated responses\n\nIn addition to end-to-end tests using real AI-generated responses, we run some end-to-end tests against test environments configured to return mock responses. This allows us to verify changes to GitLab code and components that don’t depend on responses generated by an AI model more frequently.\n\n> For a closer look at end-to-end testing, read our [end-to-end testing guide](https://docs.gitlab.com/ee/development/testing_guide/end_to_end/).\n### Exploratory testing and dogfooding\n\nAI features are built by humans for humans. At GitLab, exploratory testing and dogfooding greatly benefit us. GitLab team members are passionate about what features get shipped, and insights from internal usage are invaluable in shaping the direction of AI features.\n\n[Exploratory testing](https://about.gitlab.com/topics/devops/devops-test-automation/#test-automation-stages) allows the team to creatively exercise features to help ensure edge case bugs are identified and resolved. Dogfooding encourages team members to use AI features in their daily workflows, which helps us identify realistic issues from realistic users. For a comprehensive look at how we dogfood AI features, look at [Developing GitLab Duo: How we are dogfooding our AI features](https://about.gitlab.com/blog/developing-gitlab-duo-how-we-are-dogfooding-our-ai-features/).\n\n## Get started with GitLab Duo\nHopefully this article gives you insight into how we are validating AI features at GitLab. We have integrated our team's process into our overall development as we iterate on GitLab Duo features. We encourage you to try GitLab Duo in your organization and reap the benefits of AI-powered workflows.\n\n> Start a [free trial of GitLab Duo](https://about.gitlab.com/gitlab-duo/#free-trial) today!\n\n_Members of the GitLab Test Platform team contributed to this article._\n",[25,26,27,28,29,30],"AI/ML","features","DevSecOps platform","inside GitLab","testing","performance","yml",{},"/en-us/blog/inside-look-how-gitlabs-test-platform-team-validates-ai-features",{"title":16,"description":17,"ogTitle":16,"ogDescription":17,"noIndex":35,"ogImage":21,"ogUrl":36,"ogSiteName":37,"ogType":38,"canonicalUrls":36},false,"https://about.gitlab.com/blog/inside-look-how-gitlabs-test-platform-team-validates-ai-features","https://about.gitlab.com","article","en-us/blog/inside-look-how-gitlabs-test-platform-team-validates-ai-features",[41,26,42,43,29,30],"aiml","devsecops-platform","inside-gitlab","0iwT2zmyicvhdx9xNFot1QXdEEdWgIUfBf9JGXCIfYA",{"data":46},{"logo":47,"freeTrial":52,"sales":57,"login":62,"items":67,"search":375,"minimal":406,"duo":425,"pricingDeployment":435},{"config":48},{"href":49,"dataGaName":50,"dataGaLocation":51},"/","gitlab logo","header",{"text":53,"config":54},"Get free trial",{"href":55,"dataGaName":56,"dataGaLocation":51},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com&glm_content=default-saas-trial/","free trial",{"text":58,"config":59},"Talk to sales",{"href":60,"dataGaName":61,"dataGaLocation":51},"/sales/","sales",{"text":63,"config":64},"Sign in",{"href":65,"dataGaName":66,"dataGaLocation":51},"https://gitlab.com/users/sign_in/","sign in",[68,95,190,195,296,356],{"text":69,"config":70,"cards":72},"Platform",{"dataNavLevelOne":71},"platform",[73,79,87],{"title":69,"description":74,"link":75},"The intelligent orchestration platform for DevSecOps",{"text":76,"config":77},"Explore our Platform",{"href":78,"dataGaName":71,"dataGaLocation":51},"/platform/",{"title":80,"description":81,"link":82},"GitLab Duo Agent Platform","Agentic AI for the entire software lifecycle",{"text":83,"config":84},"Meet GitLab Duo",{"href":85,"dataGaName":86,"dataGaLocation":51},"/gitlab-duo-agent-platform/","gitlab duo agent platform",{"title":88,"description":89,"link":90},"Why GitLab","See the top reasons enterprises choose GitLab",{"text":91,"config":92},"Learn more",{"href":93,"dataGaName":94,"dataGaLocation":51},"/why-gitlab/","why gitlab",{"text":96,"left":13,"config":97,"link":99,"lists":103,"footer":172},"Product",{"dataNavLevelOne":98},"solutions",{"text":100,"config":101},"View all Solutions",{"href":102,"dataGaName":98,"dataGaLocation":51},"/solutions/",[104,128,151],{"title":105,"description":106,"link":107,"items":112},"Automation","CI/CD and automation to accelerate deployment",{"config":108},{"icon":109,"href":110,"dataGaName":111,"dataGaLocation":51},"AutomatedCodeAlt","/solutions/delivery-automation/","automated software delivery",[113,117,120,124],{"text":114,"config":115},"CI/CD",{"href":116,"dataGaLocation":51,"dataGaName":114},"/solutions/continuous-integration/",{"text":80,"config":118},{"href":85,"dataGaLocation":51,"dataGaName":119},"gitlab duo agent platform - product menu",{"text":121,"config":122},"Source Code Management",{"href":123,"dataGaLocation":51,"dataGaName":121},"/solutions/source-code-management/",{"text":125,"config":126},"Automated Software Delivery",{"href":110,"dataGaLocation":51,"dataGaName":127},"Automated software delivery",{"title":129,"description":130,"link":131,"items":136},"Security","Deliver code faster without compromising security",{"config":132},{"href":133,"dataGaName":134,"dataGaLocation":51,"icon":135},"/solutions/application-security-testing/","security and compliance","ShieldCheckLight",[137,141,146],{"text":138,"config":139},"Application Security Testing",{"href":133,"dataGaName":140,"dataGaLocation":51},"Application security testing",{"text":142,"config":143},"Software Supply Chain Security",{"href":144,"dataGaLocation":51,"dataGaName":145},"/solutions/supply-chain/","Software supply chain security",{"text":147,"config":148},"Software Compliance",{"href":149,"dataGaName":150,"dataGaLocation":51},"/solutions/software-compliance/","software compliance",{"title":152,"link":153,"items":158},"Measurement",{"config":154},{"icon":155,"href":156,"dataGaName":157,"dataGaLocation":51},"DigitalTransformation","/solutions/visibility-measurement/","visibility and measurement",[159,163,167],{"text":160,"config":161},"Visibility & Measurement",{"href":156,"dataGaLocation":51,"dataGaName":162},"Visibility and Measurement",{"text":164,"config":165},"Value Stream Management",{"href":166,"dataGaLocation":51,"dataGaName":164},"/solutions/value-stream-management/",{"text":168,"config":169},"Analytics & Insights",{"href":170,"dataGaLocation":51,"dataGaName":171},"/solutions/analytics-and-insights/","Analytics and insights",{"title":173,"items":174},"GitLab for",[175,180,185],{"text":176,"config":177},"Enterprise",{"href":178,"dataGaLocation":51,"dataGaName":179},"/enterprise/","enterprise",{"text":181,"config":182},"Small Business",{"href":183,"dataGaLocation":51,"dataGaName":184},"/small-business/","small business",{"text":186,"config":187},"Public Sector",{"href":188,"dataGaLocation":51,"dataGaName":189},"/solutions/public-sector/","public sector",{"text":191,"config":192},"Pricing",{"href":193,"dataGaName":194,"dataGaLocation":51,"dataNavLevelOne":194},"/pricing/","pricing",{"text":196,"config":197,"link":199,"lists":203,"feature":283},"Resources",{"dataNavLevelOne":198},"resources",{"text":200,"config":201},"View all resources",{"href":202,"dataGaName":198,"dataGaLocation":51},"/resources/",[204,237,255],{"title":205,"items":206},"Getting started",[207,212,217,222,227,232],{"text":208,"config":209},"Install",{"href":210,"dataGaName":211,"dataGaLocation":51},"/install/","install",{"text":213,"config":214},"Quick start guides",{"href":215,"dataGaName":216,"dataGaLocation":51},"/get-started/","quick setup checklists",{"text":218,"config":219},"Learn",{"href":220,"dataGaLocation":51,"dataGaName":221},"https://university.gitlab.com/","learn",{"text":223,"config":224},"Product documentation",{"href":225,"dataGaName":226,"dataGaLocation":51},"https://docs.gitlab.com/","product documentation",{"text":228,"config":229},"Best practice videos",{"href":230,"dataGaName":231,"dataGaLocation":51},"/getting-started-videos/","best practice videos",{"text":233,"config":234},"Integrations",{"href":235,"dataGaName":236,"dataGaLocation":51},"/integrations/","integrations",{"title":238,"items":239},"Discover",[240,245,250],{"text":241,"config":242},"Customer success stories",{"href":243,"dataGaName":244,"dataGaLocation":51},"/customers/","customer success stories",{"text":246,"config":247},"Blog",{"href":248,"dataGaName":249,"dataGaLocation":51},"/blog/","blog",{"text":251,"config":252},"Remote",{"href":253,"dataGaName":254,"dataGaLocation":51},"https://handbook.gitlab.com/handbook/company/culture/all-remote/","remote",{"title":256,"items":257},"Connect",[258,263,268,273,278],{"text":259,"config":260},"GitLab Services",{"href":261,"dataGaName":262,"dataGaLocation":51},"/services/","services",{"text":264,"config":265},"Community",{"href":266,"dataGaName":267,"dataGaLocation":51},"/community/","community",{"text":269,"config":270},"Forum",{"href":271,"dataGaName":272,"dataGaLocation":51},"https://forum.gitlab.com/","forum",{"text":274,"config":275},"Events",{"href":276,"dataGaName":277,"dataGaLocation":51},"/events/","events",{"text":279,"config":280},"Partners",{"href":281,"dataGaName":282,"dataGaLocation":51},"/partners/","partners",{"backgroundColor":284,"textColor":285,"text":286,"image":287,"link":291},"#2f2a6b","#fff","Insights for the future of software development",{"altText":288,"config":289},"the source promo card",{"src":290},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758208064/dzl0dbift9xdizyelkk4.svg",{"text":292,"config":293},"Read the latest",{"href":294,"dataGaName":295,"dataGaLocation":51},"/the-source/","the source",{"text":297,"config":298,"lists":300},"Company",{"dataNavLevelOne":299},"company",[301],{"items":302},[303,308,314,316,321,326,331,336,341,346,351],{"text":304,"config":305},"About",{"href":306,"dataGaName":307,"dataGaLocation":51},"/company/","about",{"text":309,"config":310,"footerGa":313},"Jobs",{"href":311,"dataGaName":312,"dataGaLocation":51},"/jobs/","jobs",{"dataGaName":312},{"text":274,"config":315},{"href":276,"dataGaName":277,"dataGaLocation":51},{"text":317,"config":318},"Leadership",{"href":319,"dataGaName":320,"dataGaLocation":51},"/company/team/e-group/","leadership",{"text":322,"config":323},"Team",{"href":324,"dataGaName":325,"dataGaLocation":51},"/company/team/","team",{"text":327,"config":328},"Handbook",{"href":329,"dataGaName":330,"dataGaLocation":51},"https://handbook.gitlab.com/","handbook",{"text":332,"config":333},"Investor relations",{"href":334,"dataGaName":335,"dataGaLocation":51},"https://ir.gitlab.com/","investor relations",{"text":337,"config":338},"Trust Center",{"href":339,"dataGaName":340,"dataGaLocation":51},"/security/","trust center",{"text":342,"config":343},"AI Transparency Center",{"href":344,"dataGaName":345,"dataGaLocation":51},"/ai-transparency-center/","ai transparency center",{"text":347,"config":348},"Newsletter",{"href":349,"dataGaName":350,"dataGaLocation":51},"/company/contact/#contact-forms","newsletter",{"text":352,"config":353},"Press",{"href":354,"dataGaName":355,"dataGaLocation":51},"/press/","press",{"text":357,"config":358,"lists":359},"Contact us",{"dataNavLevelOne":299},[360],{"items":361},[362,365,370],{"text":58,"config":363},{"href":60,"dataGaName":364,"dataGaLocation":51},"talk to sales",{"text":366,"config":367},"Support portal",{"href":368,"dataGaName":369,"dataGaLocation":51},"https://support.gitlab.com","support portal",{"text":371,"config":372},"Customer portal",{"href":373,"dataGaName":374,"dataGaLocation":51},"https://customers.gitlab.com/customers/sign_in/","customer portal",{"close":376,"login":377,"suggestions":384},"Close",{"text":378,"link":379},"To search repositories and projects, login to",{"text":380,"config":381},"gitlab.com",{"href":65,"dataGaName":382,"dataGaLocation":383},"search login","search",{"text":385,"default":386},"Suggestions",[387,389,393,395,399,403],{"text":80,"config":388},{"href":85,"dataGaName":80,"dataGaLocation":383},{"text":390,"config":391},"Code Suggestions (AI)",{"href":392,"dataGaName":390,"dataGaLocation":383},"/solutions/code-suggestions/",{"text":114,"config":394},{"href":116,"dataGaName":114,"dataGaLocation":383},{"text":396,"config":397},"GitLab on AWS",{"href":398,"dataGaName":396,"dataGaLocation":383},"/partners/technology-partners/aws/",{"text":400,"config":401},"GitLab on Google Cloud",{"href":402,"dataGaName":400,"dataGaLocation":383},"/partners/technology-partners/google-cloud-platform/",{"text":404,"config":405},"Why GitLab?",{"href":93,"dataGaName":404,"dataGaLocation":383},{"freeTrial":407,"mobileIcon":412,"desktopIcon":417,"secondaryButton":420},{"text":408,"config":409},"Start free trial",{"href":410,"dataGaName":56,"dataGaLocation":411},"https://gitlab.com/-/trials/new/","nav",{"altText":413,"config":414},"Gitlab Icon",{"src":415,"dataGaName":416,"dataGaLocation":411},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203874/jypbw1jx72aexsoohd7x.svg","gitlab icon",{"altText":413,"config":418},{"src":419,"dataGaName":416,"dataGaLocation":411},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203875/gs4c8p8opsgvflgkswz9.svg",{"text":421,"config":422},"Get Started",{"href":423,"dataGaName":424,"dataGaLocation":411},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com/compare/gitlab-vs-github/","get started",{"freeTrial":426,"mobileIcon":431,"desktopIcon":433},{"text":427,"config":428},"Learn more about GitLab Duo",{"href":429,"dataGaName":430,"dataGaLocation":411},"/gitlab-duo/","gitlab duo",{"altText":413,"config":432},{"src":415,"dataGaName":416,"dataGaLocation":411},{"altText":413,"config":434},{"src":419,"dataGaName":416,"dataGaLocation":411},{"freeTrial":436,"mobileIcon":441,"desktopIcon":443},{"text":437,"config":438},"Back to pricing",{"href":193,"dataGaName":439,"dataGaLocation":411,"icon":440},"back to pricing","GoBack",{"altText":413,"config":442},{"src":415,"dataGaName":416,"dataGaLocation":411},{"altText":413,"config":444},{"src":419,"dataGaName":416,"dataGaLocation":411},{"title":446,"button":447,"config":452},"See how agentic AI transforms software delivery",{"text":448,"config":449},"Watch GitLab Transcend now",{"href":450,"dataGaName":451,"dataGaLocation":51},"/events/transcend/virtual/","transcend event",{"layout":453,"icon":454},"release","AiStar",{"data":456},{"text":457,"source":458,"edit":464,"contribute":469,"config":474,"items":479,"minimal":685},"Git is a trademark of Software Freedom Conservancy and our use of 'GitLab' is under license",{"text":459,"config":460},"View page source",{"href":461,"dataGaName":462,"dataGaLocation":463},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/","page source","footer",{"text":465,"config":466},"Edit this page",{"href":467,"dataGaName":468,"dataGaLocation":463},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/content/","web ide",{"text":470,"config":471},"Please contribute",{"href":472,"dataGaName":473,"dataGaLocation":463},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/CONTRIBUTING.md/","please contribute",{"twitter":475,"facebook":476,"youtube":477,"linkedin":478},"https://twitter.com/gitlab","https://www.facebook.com/gitlab","https://www.youtube.com/channel/UCnMGQ8QHMAnVIsI3xJrihhg","https://www.linkedin.com/company/gitlab-com",[480,527,580,624,651],{"title":191,"links":481,"subMenu":496},[482,486,491],{"text":483,"config":484},"View plans",{"href":193,"dataGaName":485,"dataGaLocation":463},"view plans",{"text":487,"config":488},"Why Premium?",{"href":489,"dataGaName":490,"dataGaLocation":463},"/pricing/premium/","why premium",{"text":492,"config":493},"Why Ultimate?",{"href":494,"dataGaName":495,"dataGaLocation":463},"/pricing/ultimate/","why ultimate",[497],{"title":498,"links":499},"Contact Us",[500,503,505,507,512,517,522],{"text":501,"config":502},"Contact sales",{"href":60,"dataGaName":61,"dataGaLocation":463},{"text":366,"config":504},{"href":368,"dataGaName":369,"dataGaLocation":463},{"text":371,"config":506},{"href":373,"dataGaName":374,"dataGaLocation":463},{"text":508,"config":509},"Status",{"href":510,"dataGaName":511,"dataGaLocation":463},"https://status.gitlab.com/","status",{"text":513,"config":514},"Terms of use",{"href":515,"dataGaName":516,"dataGaLocation":463},"/terms/","terms of use",{"text":518,"config":519},"Privacy statement",{"href":520,"dataGaName":521,"dataGaLocation":463},"/privacy/","privacy statement",{"text":523,"config":524},"Cookie preferences",{"dataGaName":525,"dataGaLocation":463,"id":526,"isOneTrustButton":13},"cookie preferences","ot-sdk-btn",{"title":96,"links":528,"subMenu":536},[529,532],{"text":27,"config":530},{"href":78,"dataGaName":531,"dataGaLocation":463},"devsecops platform",{"text":533,"config":534},"AI-Assisted Development",{"href":429,"dataGaName":535,"dataGaLocation":463},"ai-assisted development",[537],{"title":538,"links":539},"Topics",[540,545,550,555,560,565,570,575],{"text":541,"config":542},"CICD",{"href":543,"dataGaName":544,"dataGaLocation":463},"/topics/ci-cd/","cicd",{"text":546,"config":547},"GitOps",{"href":548,"dataGaName":549,"dataGaLocation":463},"/topics/gitops/","gitops",{"text":551,"config":552},"DevOps",{"href":553,"dataGaName":554,"dataGaLocation":463},"/topics/devops/","devops",{"text":556,"config":557},"Version Control",{"href":558,"dataGaName":559,"dataGaLocation":463},"/topics/version-control/","version control",{"text":561,"config":562},"DevSecOps",{"href":563,"dataGaName":564,"dataGaLocation":463},"/topics/devsecops/","devsecops",{"text":566,"config":567},"Cloud Native",{"href":568,"dataGaName":569,"dataGaLocation":463},"/topics/cloud-native/","cloud native",{"text":571,"config":572},"AI for Coding",{"href":573,"dataGaName":574,"dataGaLocation":463},"/topics/devops/ai-for-coding/","ai for coding",{"text":576,"config":577},"Agentic AI",{"href":578,"dataGaName":579,"dataGaLocation":463},"/topics/agentic-ai/","agentic ai",{"title":581,"links":582},"Solutions",[583,585,587,592,596,599,603,606,608,611,614,619],{"text":138,"config":584},{"href":133,"dataGaName":138,"dataGaLocation":463},{"text":127,"config":586},{"href":110,"dataGaName":111,"dataGaLocation":463},{"text":588,"config":589},"Agile development",{"href":590,"dataGaName":591,"dataGaLocation":463},"/solutions/agile-delivery/","agile delivery",{"text":593,"config":594},"SCM",{"href":123,"dataGaName":595,"dataGaLocation":463},"source code management",{"text":541,"config":597},{"href":116,"dataGaName":598,"dataGaLocation":463},"continuous integration & delivery",{"text":600,"config":601},"Value stream management",{"href":166,"dataGaName":602,"dataGaLocation":463},"value stream management",{"text":546,"config":604},{"href":605,"dataGaName":549,"dataGaLocation":463},"/solutions/gitops/",{"text":176,"config":607},{"href":178,"dataGaName":179,"dataGaLocation":463},{"text":609,"config":610},"Small business",{"href":183,"dataGaName":184,"dataGaLocation":463},{"text":612,"config":613},"Public sector",{"href":188,"dataGaName":189,"dataGaLocation":463},{"text":615,"config":616},"Education",{"href":617,"dataGaName":618,"dataGaLocation":463},"/solutions/education/","education",{"text":620,"config":621},"Financial services",{"href":622,"dataGaName":623,"dataGaLocation":463},"/solutions/finance/","financial services",{"title":196,"links":625},[626,628,630,632,635,637,639,641,643,645,647,649],{"text":208,"config":627},{"href":210,"dataGaName":211,"dataGaLocation":463},{"text":213,"config":629},{"href":215,"dataGaName":216,"dataGaLocation":463},{"text":218,"config":631},{"href":220,"dataGaName":221,"dataGaLocation":463},{"text":223,"config":633},{"href":225,"dataGaName":634,"dataGaLocation":463},"docs",{"text":246,"config":636},{"href":248,"dataGaName":249,"dataGaLocation":463},{"text":241,"config":638},{"href":243,"dataGaName":244,"dataGaLocation":463},{"text":251,"config":640},{"href":253,"dataGaName":254,"dataGaLocation":463},{"text":259,"config":642},{"href":261,"dataGaName":262,"dataGaLocation":463},{"text":264,"config":644},{"href":266,"dataGaName":267,"dataGaLocation":463},{"text":269,"config":646},{"href":271,"dataGaName":272,"dataGaLocation":463},{"text":274,"config":648},{"href":276,"dataGaName":277,"dataGaLocation":463},{"text":279,"config":650},{"href":281,"dataGaName":282,"dataGaLocation":463},{"title":297,"links":652},[653,655,657,659,661,663,665,669,674,676,678,680],{"text":304,"config":654},{"href":306,"dataGaName":299,"dataGaLocation":463},{"text":309,"config":656},{"href":311,"dataGaName":312,"dataGaLocation":463},{"text":317,"config":658},{"href":319,"dataGaName":320,"dataGaLocation":463},{"text":322,"config":660},{"href":324,"dataGaName":325,"dataGaLocation":463},{"text":327,"config":662},{"href":329,"dataGaName":330,"dataGaLocation":463},{"text":332,"config":664},{"href":334,"dataGaName":335,"dataGaLocation":463},{"text":666,"config":667},"Sustainability",{"href":668,"dataGaName":666,"dataGaLocation":463},"/sustainability/",{"text":670,"config":671},"Diversity, inclusion and belonging (DIB)",{"href":672,"dataGaName":673,"dataGaLocation":463},"/diversity-inclusion-belonging/","Diversity, inclusion and belonging",{"text":337,"config":675},{"href":339,"dataGaName":340,"dataGaLocation":463},{"text":347,"config":677},{"href":349,"dataGaName":350,"dataGaLocation":463},{"text":352,"config":679},{"href":354,"dataGaName":355,"dataGaLocation":463},{"text":681,"config":682},"Modern Slavery Transparency Statement",{"href":683,"dataGaName":684,"dataGaLocation":463},"https://handbook.gitlab.com/handbook/legal/modern-slavery-act-transparency-statement/","modern slavery transparency statement",{"items":686},[687,690,693],{"text":688,"config":689},"Terms",{"href":515,"dataGaName":516,"dataGaLocation":463},{"text":691,"config":692},"Cookies",{"dataGaName":525,"dataGaLocation":463,"id":526,"isOneTrustButton":13},{"text":694,"config":695},"Privacy",{"href":520,"dataGaName":521,"dataGaLocation":463},[697,710],{"id":698,"title":19,"body":9,"config":699,"content":701,"description":9,"extension":31,"meta":705,"navigation":13,"path":706,"seo":707,"stem":708,"__hash__":709},"blogAuthors/en-us/blog/authors/mark-lapierre.yml",{"template":700},"BlogAuthor",{"name":19,"config":702},{"headshot":703,"ctfId":704},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749669066/Blog/Author%20Headshots/mark_lapierre.png","2Fnsk5H33npbli2fy9kMqu",{},"/en-us/blog/authors/mark-lapierre",{},"en-us/blog/authors/mark-lapierre","0HzPTjvC6yQJRoZypDDz_Ow6FkOfcD8aea81kGSFg7o",{"id":711,"title":20,"body":9,"config":712,"content":713,"description":9,"extension":31,"meta":717,"navigation":13,"path":718,"seo":719,"stem":720,"__hash__":721},"blogAuthors/en-us/blog/authors/vincy-wilson.yml",{"template":700},{"name":20,"config":714},{"headshot":715,"ctfId":716},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749669069/Blog/Author%20Headshots/vincy.jpg","1iyKndVlbE3dQnxOJoSY0q",{},"/en-us/blog/authors/vincy-wilson",{},"en-us/blog/authors/vincy-wilson","LvCuCwCbFKkkdcuAkl-tE7wn13eg9fSlEk_3q8yMPIg",[723,736,749],{"content":724,"config":734},{"title":725,"description":726,"authors":727,"heroImage":729,"date":730,"body":731,"category":10,"tags":732},"10 AI prompts to speed your team’s software delivery","Eliminate review backlogs, security delays, and coordination overhead with ready-to-use AI prompts covering every stage of the software lifecycle.",[728],"Chandler Gibbons","https://res.cloudinary.com/about-gitlab-com/image/upload/v1772632341/duj8vaznbhtyxxhodb17.png","2026-03-04","AI-assisted coding tools are helping developers generate code faster than ever. So why aren’t teams _shipping_ faster?\n\nBecause coding is only 20% of the software delivery lifecycle, the remaining 80% becomes the bottleneck: code review backlogs grow, security scanning can’t keep pace, documentation falls behind, and manual coordination overhead increases.\n\nThe good news is that the same AI capabilities that accelerate individual coding can eliminate these team-level delays. You just need to apply AI across your entire software lifecycle, not only during the coding phase.\n\nBelow are 10 ready-to-use prompts from the [GitLab Duo Agent Platform Prompt Library](https://about.gitlab.com/gitlab-duo/prompt-library/) that help teams overcome common obstacles to faster software delivery. Each prompt addresses a specific slowdown that emerges when individual productivity increases without corresponding improvements in team processes.\n\n## How do you move code review from bottleneck to accelerator?\nDevelopers generate merge requests faster with AI assistance, but human reviewers can quickly become overwhelmed as code review cycles stretch from hours to days. AI can handle routine review tasks, freeing reviewers to focus on architecture and business logic instead of catching basic logical errors and API contract violations.\n\n### Review MR for logical errors\n**Complexity**: Beginner\n\n**Category**: Code Review\n\n**Prompt from library**:\n\n\n```text\nReview this MR for logical errors, edge cases, and potential bugs: [MR URL or paste code]\n```\n\n**Why it helps**: Automated linters catch syntax issues, but logical errors require understanding intent. This prompt catches bugs before human reviewers even look at the code, reducing review cycles from multiple rounds to often just one approval.\n\n### Identify breaking changes in MR\n**Complexity**: Beginner\n\n**Category**: Code Review\n\n**Prompt from library**:\n\n\n```text\nDoes this MR introduce any breaking changes?\n\nChanges:\n[PASTE CODE DIFF]\n\nCheck for:\n1. API signature changes\n2. Removed or renamed public methods\n3. Changed return types\n4. Modified database schemas\n5. Breaking configuration changes\n```\n\n**Why it helps**: Breaking changes discovered during deployment can cause rollbacks and incidents. This prompt shifts that discovery left to the MR stage, when fixes are faster and less expensive.\n\n## How can you shift security left without slowing down?\nSecurity scans generate hundreds of findings. Security teams manually triage each one while developers wait for approval to deploy. Most findings are false positives or low-risk issues, but identifying the real threats requires expertise and time. AI can prioritize findings by actual exploitability and auto-remediate common vulnerabilities, allowing security teams to focus on the threats that matter.\n\n### Analyze security scan results\n**Complexity**: Intermediate\n\n**Category**: Security\n\n**Agent**: Duo Security Analyst\n\n**Prompt from library**:\n\n\n```text\n@security_analyst Analyze these security scan results:\n\n[PASTE SCAN OUTPUT]\n\nFor each finding:\n1. Assess real risk vs false positive\n2. Explain the vulnerability\n3. Suggest remediation\n4. Prioritize by severity\n```\n\n**Why it helps**: Most security scan findings are false positives or low-risk issues. This prompt helps security teams focus on the findings that actually matter, reducing remediation time from weeks to days.\n\n### Review code for security issues\n**Complexity**: Intermediate\n\n**Category**: Security\n\n**Agent**: Duo Security Analyst\n\n**Prompt from library**:\n\n```text\n@security_analyst Review this code for security issues:\n\n[PASTE CODE]\n\nCheck for:\n1. Injection vulnerabilities\n2. Authentication/authorization flaws\n3. Data exposure risks\n4. Insecure dependencies\n5. Cryptographic issues\n```\n\n**Why it helps**: Traditional security reviews happen after code is written. This prompt enables developers to find and fix security issues before creating an MR, eliminating the back and forth that delays deployments.\n\n## How do you keep documentation current as code changes?\nCode changes faster than documentation. Onboarding new developers takes weeks because docs are outdated or missing. Teams know documentation is important, but it always gets deferred when deadlines approach. Automating documentation generation and updates as part of your standard workflow ensures docs stay current without adding manual work.\n\n### Generate release notes from MRs\n**Complexity**: Beginner\n\n**Category**: Documentation\n\n**Prompt from library**:\n\n```text\nGenerate release notes for these merged MRs:\n[LIST MR URLs or paste titles]\n\nGroup by:\n1. New features\n2. Bug fixes\n3. Performance improvements\n4. Breaking changes\n5. Deprecations\n```\n\n**Why it helps**: Manual release note compilation takes hours and often includes errors or omissions. Automated generation ensures every release has comprehensive notes without adding work to your release process.\n\n### Update documentation after code changes\n**Complexity**: Beginner\n\n**Category**: Documentation\n\n**Prompt from library**:\n\n```text\nI changed this code:\n\n[PASTE CODE CHANGES]\n\nWhat documentation needs updating? Check:\n1. README files\n2. API documentation\n3. Architecture diagrams\n4. Onboarding guides\n```\n\n**Why it helps**: Documentation drift happens because teams forget which docs need updates after code changes. This prompt makes documentation maintenance part of your development workflow, not a separate task that gets deferred.\n\n## How do you break down planning complexity?\nLarge features get stuck in planning. Teams spend weeks in meetings trying to scope work and identify dependencies. The complexity feels overwhelming, and it's hard to know where to start. AI can systematically decompose complex work into concrete, implementable tasks with clear dependencies and acceptance criteria, transforming weeks of planning into focused implementation.\n\n### Break down epic into issues\n**Complexity**: Intermediate\n\n**Category**: Documentation\n\n**Agent**: Duo Planner\n\n**Prompt from library**:\n\n```text\nBreak down this epic into implementable issues:\n\n[EPIC DESCRIPTION]\n\nConsider:\n1. Technical dependencies\n2. Reasonable issue sizes\n3. Clear acceptance criteria\n4. Logical implementation order\n```\n\n**Why it helps**: This prompt transforms a week of planning meetings into 30 minutes of AI-assisted decomposition followed by team review. Teams start implementation sooner with clearer direction.\n\n## How can you expand test coverage without expanding effort?\nDevelopers are writing code faster, but if testing doesn't keep pace, test coverage decreases and bugs slip through. Writing comprehensive tests manually is time-consuming, and developers often miss edge cases under deadline pressure. Generating tests automatically means developers can review and refine rather than write from scratch, maintaining quality without sacrificing velocity.\n\n### Generate unit tests\n**Complexity**: Beginner\n\n**Category**: Testing\n\n**Prompt from library**:\n\n```text\nGenerate unit tests for this function:\n\n[PASTE FUNCTION]\n\nInclude tests for:\n1. Happy path\n2. Edge cases\n3. Error conditions\n4. Boundary values\n5. Invalid inputs\n```\n\n**Why it helps**: Writing tests manually is time consuming, and developers often miss edge cases. This prompt generates thorough test suites in seconds, which developers can review and adjust rather than write from scratch.\n\n### Review test coverage gaps\n**Complexity**: Beginner\n\n**Category**: Testing\n\n**Prompt from library**:\n\n```text\nAnalyze test coverage for [MODULE/COMPONENT]:\n\nCurrent coverage: [PERCENTAGE]\n\nIdentify:\n1. Untested functions/methods\n2. Uncovered edge cases\n3. Missing error scenario tests\n4. Integration points without tests\n5. Priority areas to test next\n```\n\n**Why it helps**: This prompt reveals blind spots in your test suite before they cause production incidents. Teams can systematically improve coverage where it matters most.\n\n## How do you reduce mean time to resolution when debugging?\nProduction incidents take hours to diagnose. Developers wade through logs and stack traces while customers experience downtime. Every minute of debugging is a minute of lost productivity and potential revenue. AI can accelerate root cause analysis by parsing complex error messages and suggesting specific fixes, cutting diagnostic time from hours to minutes.\n\n### Debug failing pipeline\n**Complexity**: Beginner\n\n**Category**: Debugging\n\n**Prompt from library**:\n\n```text\nThis pipeline is failing:\n\nJob: [JOB NAME]\nStage: [STAGE]\nError: [PASTE ERROR MESSAGE/LOG]\n\nHelp me:\n1. Identify the root cause\n2. Suggest a fix\n3. Explain why it started failing\n4. Prevent similar issues\n```\n\n**Why it helps**: CI/CD failures block entire teams. This prompt diagnoses failures in seconds instead of the 15-30 minutes developers typically spend investigating, keeping deployment velocity high.\n\n## Moving from individual gains to team acceleration\nThese prompts represent a shift in how teams apply AI to software delivery. Rather than focusing solely on individual developer productivity, they address the coordination, quality, and knowledge-sharing challenges that actually constrain team velocity.\n\nThe [complete prompt library](https://about.gitlab.com/gitlab-duo/prompt-library/) contains more than 100 prompts across all stages of the software lifecycle: planning, development, security, testing, deployment, and operations. Each prompt is tagged by complexity level (Beginner, Intermediate, Advanced) and categorized by use case, making it easy to find the right starting point for your team.\n\nStart with prompts tagged “Beginner” that address your team’s most pressing obstacles. As your team builds confidence, explore intermediate and advanced prompts that enable more sophisticated workflows. The goal is not just faster coding — it's faster, safer, higher-quality software delivery from planning through production.",[25,733],"DevOps platform",{"featured":35,"template":14,"slug":735},"10-ai-prompts-to-speed-your-teams-software-delivery",{"content":737,"config":747},{"title":738,"description":739,"heroImage":740,"authors":741,"date":743,"body":744,"category":10,"tags":745},"AI can detect vulnerabilities, but who governs risk?","AI-assisted vulnerability detection is developing fast, but the harder challenges of enforcement, governance, and supply chain security require a holistic platform.","https://res.cloudinary.com/about-gitlab-com/image/upload/v1772195014/ooezwusxjl1f7ijfmbvj.png",[742],"Omer Azaria","2026-02-27","Anthropic recently announced Claude Code Security, an AI system that detects vulnerabilities and proposes fixes. The market reacted immediately, with security stocks dipping as investors questioned whether AI might replace traditional AppSec tools. The question on everyone's mind: If AI can write code and secure it, is application security about to become obsolete?\n\nIf security only meant scanning code, the answer might be yes. But enterprise security has never been about detection alone.\n\nOrganizations are not asking whether AI can find vulnerabilities. They are asking three much harder questions: \n\n* Is what we are about to ship safe?  \n* Has our risk posture changed as environments evolve and dependencies, third-party services, tools, and infrastructure continuously shift?  \n* How do we govern a codebase that is increasingly assembled by AI and third-party sources, and that we are still accountable for? \n\nThose questions require a platform answer: Detection surfaces risk, but governance determines what happens next. \n\n[GitLab](https://about.gitlab.com/) is the orchestration layer built to govern the software lifecycle end-to-end. It gives teams the enforcement, visibility, and auditability they need to keep pace with the speed of AI-assisted development.\n\n## Trusting AI requires governing risk\n\nAI systems are rapidly getting better at identifying vulnerabilities and suggesting fixes. This is a meaningful and welcome advancement, but analysis is not accountability.\n\nAI cannot enforce company policy or define acceptable risk on its own. Humans must set the boundaries, policies, and guardrails that agents operate within, establishing separation of duties, ensuring audit trails, and maintaining consistent controls across thousands of repositories and teams. Trust in agents comes not from autonomy alone, but from clearly defined governance set by people. \n\nIn an [agentic world](https://about.gitlab.com/topics/agentic-ai/), where software is increasingly written and modified by autonomous systems, governance becomes more important, not less. The more autonomy organizations grant to AI, the stronger the governance must be.\n\nGovernance is not friction. It is the foundation that makes AI-assisted development trustworthy at scale.\n\n## LLMs see code, but platforms see context\n\nA large language model ([LLM](https://about.gitlab.com/blog/what-is-a-large-language-model-llm/)) evaluates code in isolation. An enterprise application security platform understands context. This difference matters because risk decisions are contextual:\n\n* Who authored the change?  \n* How critical is the application to the business?  \n* How does it interact with infrastructure and dependencies?  \n* Does the vulnerability exist in code that is actually reachable in production, or is it buried in a dependency that never executes?  \n* Is it actually exploitable in production, given how the application runs, its APIs, and the environment around it?\n\nSecurity decisions depend on this context. Without it, detection produces noisy alerts that slow down development rather than reducing risk. With it, organizations can triage quickly and manage risk effectively. Context evolves continuously as software changes, which means governance cannot be a one-time decision. \n\n## Static scans can’t keep up with dynamic risk\n\nSoftware risk is dynamic. Dependencies change, environments evolve, and systems interact in ways no single analysis can fully predict. A clean scan at one moment does not guarantee safety at release.\n\nEnterprise security depends on continuous assurance: controls embedded directly into development workflows that evaluate risk as software is built, tested, and deployed.\n\nDetection provides insight. Governance provides trust. Continuous governance is what allows organizations to ship safely at scale.\n\n## Governing the agentic future\n\nAI is reshaping how software is created. The question is no longer whether teams will use AI, but how safely they can scale it.\n\nSoftware today is assembled as much as it is written, from AI-generated code, open-source libraries, and third-party dependencies that span thousands of projects. Governing what ships across all of those sources is the hardest and most consequential part of application security, and it is the part that no developer-side tool is built to address. \n\nAs an intelligent orchestration platform, GitLab is built to address this problem. GitLab Ultimate embeds governance, policy enforcement, security scanning, and auditability directly into the workflows where software is planned, built, and shipped, so security teams can govern at the speed of AI. \n\nAI will accelerate development dramatically. The organizations that benefit most from AI will not be those with the smartest assistants alone, but those that build trust through strong governance.\n\n> To learn how GitLab helps organizations [govern and ship AI-generated code](https://about.gitlab.com/solutions/software-compliance/?utm_medium=blog&utm_campaign=eg_global_x_x_security_en_) safely, [talk to our team today](https://about.gitlab.com/sales/?utm_medium=blog&utm_campaign=eg_global_x_x_security_en_)\n\n\n ## Related reading\n\n - [Integrating AI with DevOps for enhanced security](https://about.gitlab.com/topics/devops/ai-enhanced-security/)\n - [The GitLab AI Security Framework for security leaders](https://about.gitlab.com/blog/the-gitlab-ai-security-framework-for-security-leaders/)\n - [Improve AI security in GitLab with composite identities](https://about.gitlab.com/blog/improve-ai-security-in-gitlab-with-composite-identities/)",[25,746],"security",{"featured":13,"template":14,"slug":748},"ai-can-detect-vulnerabilities-but-who-governs-risk",{"content":750,"config":760},{"title":751,"description":752,"authors":753,"category":10,"tags":755,"date":757,"heroImage":758,"body":759},"Secure and fast deployments to Google Agent Engine with GitLab","Follow this step-by-step guide to build an AI agent with Google's Agent Development Kit and deploy to Agent Engine using GitLab.",[754],"Regnard Raquedan",[25,756,114,561],"google","2026-02-26","https://res.cloudinary.com/about-gitlab-com/image/upload/v1772111172/mwhgbjawn62kymfwrhle.png","In this tutorial, you'll learn how to deploy an AI agent built with Google's Agent Development Kit ([ADK](https://google.github.io/adk-docs/)) to [Agent Engine](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/overview) using GitLab's native Google Cloud integration and CI/CD pipelines. We'll cover IAM configuration, pipeline setup, and testing your deployed agent.\n\n## What is Agent Engine and why does it matter?\n\nAgent Engine is Google Cloud's managed runtime specifically designed for AI agents. Think of it as the production home for your agents — where they live, run, and scale without you having to manage the underlying infrastructure. Agent Engine handles infrastructure, scaling, session management, and memory storage so you can focus on building your agent — not managing servers. It also integrates natively with Google Cloud's logging, monitoring, and IAM.\n\n## Why use GitLab to deploy to Agent Engine?\n\nAI agent deployment is typically difficult to configure correctly. Security considerations, CI/CD orchestration, and cloud permissions create friction that slows down development cycles.\n\nGitLab streamlines this entire process while enhancing security:\n\n- **Built-in security scanning** — Every deployment is automatically scanned for vulnerabilities without additional configuration.\n- **Native Google Cloud integration** — Workload Identity Federation eliminates the need for service account keys.\n- **Simplified CI/CD** — GitLab's templates handle complex deployment logic.\n\n## Prerequisites\n\nBefore you begin, ensure you have:\n\n- A Google Cloud project with the following APIs enabled:\n  - Cloud Storage API\n  - Vertex AI API\n- A GitLab project for your source code and CI/CD pipeline\n- A Google Cloud Storage bucket for staging deployments\n- Google Cloud IAM integration configured in GitLab (see Step 1)\n\nHere are the steps to follow.\n\n## 1. Configure IAM integration\n\nThe foundation of secure deployment is proper IAM configuration between GitLab and Google Cloud using Workload Identity Federation.\n\nIn your GitLab project:\n\n1. Navigate to **Settings > Integrations**.\n2. Locate the **Google Cloud IAM** integration.\n3. Provide the following information:\n   - **Project ID**: Your Google Cloud project ID\n   - **Project Number**: Found in your Google Cloud console\n   - **Workload Identity Pool ID**: A unique identifier for your identity pool\n   - **Provider ID**: A unique identifier for your identity provider\n\nGitLab generates a script for you. Copy and run this script in Google Cloud Shell to establish the Workload Identity Federation between platforms.\n\n**Important:** Add these additional roles to your service principal for Agent Engine deployment:\n\n- `roles/aiplatform.user`\n- `roles/storage.objectAdmin`\n\nYou can add these roles using gcloud commands:\n\n```bash\nGCP_PROJECT_ID=\"\u003Cyour-project-id>\"\nGCP_PROJECT_NUMBER=\"\u003Cyour-project-number>\"\nGCP_WORKLOAD_IDENTITY_POOL=\"\u003Cyour-pool-id>\"\n\ngcloud projects add-iam-policy-binding ${GCP_PROJECT_ID} \\\n  --member=\"principalSet://iam.googleapis.com/projects/${GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${GCP_WORKLOAD_IDENTITY_POOL}/attribute.developer_access/true\" \\\n  --role='roles/aiplatform.user'\n\ngcloud projects add-iam-policy-binding ${GCP_PROJECT_ID} \\\n  --member=\"principalSet://iam.googleapis.com/projects/${GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${GCP_WORKLOAD_IDENTITY_POOL}/attribute.developer_access/true\" \\\n  --role='roles/storage.objectAdmin'\n```\n\n## 2. Create the CI/CD pipeline\n\nNow for the core of the deployment — the CI/CD pipeline. Create a `.gitlab-ci.yml` file in your project root:\n\n```yaml\nstages:\n  - test\n  - deploy\n\ncache:\n  paths:\n    - .cache/pip\n  key: ${CI_COMMIT_REF_SLUG}\n\nvariables:\n  GCP_PROJECT_ID: \"\u003Cyour-project-id>\"\n  GCP_REGION: \"us-central1\"\n  STORAGE_BUCKET: \"\u003Cyour-staging-bucket>\"\n  AGENT_NAME: \"Canada City Advisor\"\n  AGENT_ENTRY: \"canada_city_advisor\"\n\nimage: google/cloud-sdk:slim\n\n# Security scanning templates\ninclude:\n  - template: Jobs/Dependency-Scanning.gitlab-ci.yml\n  - template: Jobs/SAST.gitlab-ci.yml\n  - template: Jobs/Secret-Detection.gitlab-ci.yml\n\ndeploy-agent:\n  stage: deploy\n  identity: google_cloud\n  rules:\n    - if: $CI_COMMIT_BRANCH == \"main\"\n  before_script:\n    - gcloud config set core/disable_usage_reporting true\n    - gcloud config set component_manager/disable_update_check true\n    - pip install -q --no-cache-dir --upgrade pip google-genai google-cloud-aiplatform -r requirements.txt --break-system-packages\n  script:\n    - gcloud config set project $GCP_PROJECT_ID\n    - adk deploy agent_engine \n        --project=$GCP_PROJECT_ID \n        --region=$GCP_REGION \n        --staging_bucket=gs://$STORAGE_BUCKET \n        --display_name=\"$AGENT_NAME\" \n        $AGENT_ENTRY\n```\n\nThe pipeline consists of two stages:\n\n**Test stage** — GitLab's security scanners run automatically. The included templates provide dependency scanning, static application security testing (SAST), and secret detection without additional configuration.\n\n**Deploy stage** — Uses the ADK CLI to deploy your agent directly to Agent Engine. The staging bucket temporarily holds your application workload before Agent Engine picks it up for deployment.\n\n### Key configuration notes\n\n- The `identity: google_cloud` directive enables keyless authentication via Workload Identity Federation.\n- Security scanners are included as templates, meaning they run by default with no setup required.\n- The `adk deploy agent_engine` command handles all the complexity of packaging and deploying your agent.\n- Pipeline caching speeds up subsequent deployments by preserving pip dependencies.\n\n## 3. Deploy and verify\n\nWith your pipeline configured:\n\n1. Commit your agent code and `.gitlab-ci.yml` to GitLab.\n2. Navigate to **Build > Pipelines** to monitor execution.\n3. Watch the test stage complete security scans.\n4. Observe the deploy stage push your agent to Agent Engine.\n\nOnce the pipeline succeeds, verify your deployment in the Google Cloud Console:\n\n1. Navigate to **Vertex AI > Agent Engine**.\n2. Locate your deployed agent.\n3. Note the **resource name** — you'll need this for testing.\n\n## 4. Test your deployed agent\n\nTest your agent using a curl command. You'll need three pieces of information:\n\n- **Agent ID**: From the Agent Engine console (the resource name's numeric identifier)\n- **Project ID**: Your Google Cloud project\n- **Location**: The region where you deployed (e.g., `us-central1`)\n\n```bash\nPROJECT_ID=\"\u003Cyour-project-id>\"\nLOCATION=\"us-central1\"\nAGENT_ID=\"\u003Cyour-agent-id>\"\nTOKEN=$(gcloud auth print-access-token)\n\ncurl -X POST \\\n  -H \"Authorization: Bearer $TOKEN\" \\\n  -H \"Content-Type: application/json\" \\\n  \"https://${LOCATION}-aiplatform.googleapis.com/v1/projects/${PROJECT_ID}/locations/${LOCATION}/reasoningEngines/${AGENT_ID}:streamQuery\" \\\n  -d '{\n    \"input\": {\n      \"message\": \"I make $85,000 per year and I prefer cities with mild winters and a vibrant cultural scene. I also want to be near the coast if possible. What Canadian cities would you recommend?\",\n      \"user_id\": \"demo-user\"\n    }\n  }' | jq -r '.content.parts[0].text'\n```\n\nIf everything is configured correctly, your agent will respond with personalized city recommendations based on the budget and lifestyle preferences provided.\n\n## Security benefits of this approach\n\nThis deployment pattern provides several security advantages:\n\n- **No long-lived credentials**: Workload Identity Federation eliminates service account keys entirely.\n- **Automated vulnerability scanning**: Every deployment is scanned before reaching production.\n- **Complete audit trail**: GitLab maintains full visibility of who deployed what and when.\n- **Principle of least privilege**: Fine-grained IAM roles limit access to only what's needed.\n\n## Summary\n\nDeploying AI agents to production doesn't have to be complex. By combining GitLab's DevSecOps platform with Google Cloud's Agent Engine, you get:\n\n- A managed runtime that handles scaling and infrastructure\n- Built-in security scanning without additional tooling\n- Keyless authentication via native cloud integration\n- A streamlined deployment process that fits modern AI development workflows\n\nWatch the full demo:\n\n\n\u003Cfigure class=\"video_container\"> \u003Ciframe src=\"https://www.youtube.com/embed/sxVFa2Mk-x4?si=Oi3cUjhgd7FT2yEd\" frameborder=\"0\" allowfullscreen=\"true\" title=\"Deploy AI Agents to Agent Engine with GitLab\"> \u003C/iframe> \u003C/figure>\n\n> Ready to try it yourself? Use this tutorial's [complete code example](https://gitlab.com/gitlab-partners-public/google-cloud/demos/agent-engine-demo) to get started now. Not a GitLab customer yet? Explore the DevSecOps platform with [a free trial](https://about.gitlab.com/free-trial/).\n",{"featured":35,"template":14,"slug":761},"secure-and-fast-deployments-to-google-agent-engine-with-gitlab",{"promotions":763},[764,777,789],{"id":765,"categories":766,"header":767,"text":768,"button":769,"image":774},"ai-modernization",[10],"Is AI achieving its promise at scale?","Quiz will take 5 minutes or less",{"text":770,"config":771},"Get your AI maturity score",{"href":772,"dataGaName":773,"dataGaLocation":249},"/assessments/ai-modernization-assessment/","modernization assessment",{"config":775},{"src":776},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/qix0m7kwnd8x2fh1zq49.png",{"id":778,"categories":779,"header":781,"text":768,"button":782,"image":786},"devops-modernization",[780,564],"product","Are you just managing tools or shipping innovation?",{"text":783,"config":784},"Get your DevOps maturity score",{"href":785,"dataGaName":773,"dataGaLocation":249},"/assessments/devops-modernization-assessment/",{"config":787},{"src":788},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138785/eg818fmakweyuznttgid.png",{"id":790,"categories":791,"header":792,"text":768,"button":793,"image":797},"security-modernization",[746],"Are you trading speed for security?",{"text":794,"config":795},"Get your security maturity score",{"href":796,"dataGaName":773,"dataGaLocation":249},"/assessments/security-modernization-assessment/",{"config":798},{"src":799},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/p4pbqd9nnjejg5ds6mdk.png",{"header":801,"blurb":802,"button":803,"secondaryButton":808},"Start building faster today","See what your team can do with the intelligent orchestration platform for DevSecOps.\n",{"text":804,"config":805},"Get your free trial",{"href":806,"dataGaName":56,"dataGaLocation":807},"https://gitlab.com/-/trial_registrations/new?glm_content=default-saas-trial&glm_source=about.gitlab.com/","feature",{"text":501,"config":809},{"href":60,"dataGaName":61,"dataGaLocation":807},1772652070665]