GitLab Launches Agent-based DevSecOps Utilizing Amazon Bedrock
GitLab has expanded its partnership with Amazon Web Services, launching agent-based DevSecOps leveraging Amazon Bedrock. This enables AWS users to advance AI agent-driven development while utilizing existing cloud environments and contracts, ensuring governance and auditability.
📋 Article Processing Timeline
- 📰 Published: April 27, 2026 at 20:00
- 🔍 Collected: April 27, 2026 at 11:31
- 🤖 AI Analyzed: April 28, 2026 at 02:47 (15h 15m after Collected)
GitLab Launches Agent-based DevSecOps Utilizing Amazon Bedrock
- Routing GitLab Duo Agent Platform inferences directly to Amazon Bedrock models already running in the user's AWS account.
- GitLab credits purchased via AWS Marketplace can be included in existing AWS contracts.
- BYOM (Bring Your Own Model) functionality for Self-Managed environments allows direct connection of self-hosted AI gateways to Amazon Bedrock, keeping inference traffic and source code within each company's AWS environment.
GitLab (Headquarters: San Francisco, USA, NASDAQ: GTLB), a provider of an intelligent orchestration platform for DevSecOps, announced the expansion of its partnership with Amazon Web Services (AWS). This partnership enables AWS and GitLab users to route GitLab Duo Agent Platform inferences via Amazon Bedrock, allowing for agent-based DevSecOps while leveraging existing Amazon Bedrock models, IAM policies, and AWS contract amounts.
As AI agents generate more code, merge requests, and pipeline activities, organizations need to ensure governance and auditability at a pace that matches this increase. It is crucial to achieve this within existing environments without building new infrastructure stacks in parallel. With this expanded partnership, companies that adopt AWS as their standard environment can integrate the GitLab Duo Agent Platform seamlessly into their existing infrastructure. Furthermore, elements such as approved Amazon Bedrock models, IAM-based access control, and inclusion in AWS contract amounts can all be utilized as an extension of their current environment. This eliminates the need for new vendor risk assessments, separate model endpoint provisioning, and additional billing management.
While Amazon Bedrock provides model-level governance, GitLab's orchestration layer offers workflow-level governance in addition to the model-level governance provided by Amazon Bedrock. Specifically, policies can restrict which models agents can use, and all agent actions and generated code are recorded in audit logs. Moreover, agents operate consistently across different projects and regions through approved configurations.
This expanded partnership builds upon the recently released Bring Your Own Model (BYOM) functionality for GitLab Self-Managed users. BYOM allows direct routing of inferences from a self-hosted AI gateway to Amazon Bedrock within the user's own AWS environment, ensuring that source code and inference traffic never leave the user's network boundary. Users can leverage models fine-tuned on Amazon Bedrock directly or utilize GitLab-managed models, including the latest models from Anthropic. Regardless of the choice, GitLab handles orchestration and governance.
The GitLab Duo Agent Platform is billed per request from the total shared GitLab credits within an organization, with no per-seat fees. For users who contract GitLab via AWS Marketplace, GitLab credits are directly deducted from their existing AWS contract amount, allowing them to adjust usage within their current contract when expanding AI agents.
More details on this expanded partnership can be found here.
Comments from Stakeholders
Yoshiaki Matsuda, COO, Fixstars Corporation
"For Fixstars, which has adopted Amazon Bedrock as the standard foundation for generative AI, the GitLab Duo Agent Platform with Bring Your Own Model is highly attractive because it allows AI to be applied across the entire software delivery lifecycle while keeping inference traffic and code within the AWS environment. At Fixstars, we position AI as the core of modern software development, and with Fixstars AIBooster, we aim to improve both the quality and speed of application development."
- Routing GitLab Duo Agent Platform inferences directly to Amazon Bedrock models already running in the user's AWS account.
- GitLab credits purchased via AWS Marketplace can be included in existing AWS contracts.
- BYOM (Bring Your Own Model) functionality for Self-Managed environments allows direct connection of self-hosted AI gateways to Amazon Bedrock, keeping inference traffic and source code within each company's AWS environment.
GitLab (Headquarters: San Francisco, USA, NASDAQ: GTLB), a provider of an intelligent orchestration platform for DevSecOps, announced the expansion of its partnership with Amazon Web Services (AWS). This partnership enables AWS and GitLab users to route GitLab Duo Agent Platform inferences via Amazon Bedrock, allowing for agent-based DevSecOps while leveraging existing Amazon Bedrock models, IAM policies, and AWS contract amounts.
As AI agents generate more code, merge requests, and pipeline activities, organizations need to ensure governance and auditability at a pace that matches this increase. It is crucial to achieve this within existing environments without building new infrastructure stacks in parallel. With this expanded partnership, companies that adopt AWS as their standard environment can integrate the GitLab Duo Agent Platform seamlessly into their existing infrastructure. Furthermore, elements such as approved Amazon Bedrock models, IAM-based access control, and inclusion in AWS contract amounts can all be utilized as an extension of their current environment. This eliminates the need for new vendor risk assessments, separate model endpoint provisioning, and additional billing management.
While Amazon Bedrock provides model-level governance, GitLab's orchestration layer offers workflow-level governance in addition to the model-level governance provided by Amazon Bedrock. Specifically, policies can restrict which models agents can use, and all agent actions and generated code are recorded in audit logs. Moreover, agents operate consistently across different projects and regions through approved configurations.
This expanded partnership builds upon the recently released Bring Your Own Model (BYOM) functionality for GitLab Self-Managed users. BYOM allows direct routing of inferences from a self-hosted AI gateway to Amazon Bedrock within the user's own AWS environment, ensuring that source code and inference traffic never leave the user's network boundary. Users can leverage models fine-tuned on Amazon Bedrock directly or utilize GitLab-managed models, including the latest models from Anthropic. Regardless of the choice, GitLab handles orchestration and governance.
The GitLab Duo Agent Platform is billed per request from the total shared GitLab credits within an organization, with no per-seat fees. For users who contract GitLab via AWS Marketplace, GitLab credits are directly deducted from their existing AWS contract amount, allowing them to adjust usage within their current contract when expanding AI agents.
More details on this expanded partnership can be found here.
Comments from Stakeholders
Yoshiaki Matsuda, COO, Fixstars Corporation
"For Fixstars, which has adopted Amazon Bedrock as the standard foundation for generative AI, the GitLab Duo Agent Platform with Bring Your Own Model is highly attractive because it allows AI to be applied across the entire software delivery lifecycle while keeping inference traffic and code within the AWS environment. At Fixstars, we position AI as the core of modern software development, and with Fixstars AIBooster, we aim to improve both the quality and speed of application development."