FY26 Duo Workflow investment themes and Road-Map

Note-Work In Progress

Overview

The Workflow team is focused on enhancing Agentic Workflows within GitLab. In FY26, we will drive innovation in automation, observability, and control for AI-assisted workflows.

2025 is a crucial year for GitLab to capture the market opportunity for AI Agents and Agentic Workflows. Our goal is to position GitLab as a leader in this category.

Our narrative

The Workflow team aims to accelerate GitLab adoption of Agentic Workflows by

  • Demonstrating that lovable products can be built using the Workflow platform
  • Supporting other teams to create their Agentic Workflows to enhance their product stages

To achieve that we aim to start by working with an internal team to create and introduce a new workflow into the GitLab offering, following that we want to enable multiple teams to achieve it, without the need of close "hand-holding". By the end of the year, we want to be able to serve customers with the ability to create their own Workflows in the GitLab platform.

Vision & Strategy

  • Balancing Innovation and Governance:
    • Embrace the rapid evolution of AI capabilities while ensuring that human oversight and accountability remain integral.
    • Position Duo Workflow as a leader in responsible AI deployment—driving productivity without sacrificing compliance or quality.
  • Empowering Organizations:
    • Create a framework that serves both agile mid-sized companies and complex enterprises.
    • Enable organizations to integrate advanced AI into their software development lifecycle (SDLC) without relinquishing control over critical decisions.

Considerations

  • Controlled Agetic Workflows:
    • Empower AI agents to execute routine and complex tasks while embedding key checkpoints for human review & feedback.
    • Integrate mechanisms that will enable system level control, observability and permissions management
  • Human-in-the-Loop (HITL) Integration:
    • Ensure that every Agent key decision passes through a human review layer, guaranteeing that accountability is maintained.
    • Use explainable AI (XAI) techniques to demystify automated decisions- Show the agents plan, enable users to track and influence execution as well increased visibility to context used by the Agent.
  • Transparent Audit Trails:
    • Maintain detailed logs of AI actions to create an auditable record, enhancing trust and compliance.
    • Support the Definition ot clear roles for accountability so that, in any decision-making process, the final responsibility rests with designated human operators.
  • Drive business synergies
    • Aim to support business synergies that will enable GitLab to capture the full market opportunity

Investment Themes

1. Enhanced Agentic Workflow Productivity

  • Why it matters:
    • Agentic coding is revolutionizing developer efficiency. By improving Workflow capabilities, we reduce friction in software development and improve time-to-value.
    • Agentic Workflows can apply across the SDLC and empower more user personas such as EM's and management, Product and Project Managers, and DevOps and Platform engineers.
  • Business impact: Increased adoption of Agentic Workflows will drive revenue by enhancing GitLab’s offerings and strengthening customer retention.

2. Expansion of Agentic workflows in GitLab offerings

  • Why it matters: Developers, DevOps, Product and Project managers, EM's and leadership functions work across multiple environments, and broader support increases Workflow’s reach and adoption.
  • Business impact: Expanding Workflow across Plan capabilities, MR's, , CI/CD and more GitLab-native features enhances GitLab’s AIgentic ecosystem and drives higher adoption. By empowering internal GitLab teams to leverage Agentic Workflows in their product stages, we can leverage the impact of Agentic Workflow across the entire GitLab offering

3. Observability & Compliance Controls

  • Why it matters: Organizations need visibility into Agentic Workflow execution and the ability to enforce policies around workflow execution. Agentic workflow can be applied across the entire SDLC, Which highlights the importance of providing organizations with enterprise grade controls and observability
  • Business impact: Strengthening observability and compliance drives enterprise adoption, a key revenue segment for GitLab.

Agentic Workflow User experience

We want our user to experience:

Trust and Confidence

  • Goal: The user wants to feel certain the system won’t freeze or lose their work. They should sense that Duo Workflow is reliable and error-free.
  • Why: When users trust the system’s responsiveness, they’re more willing to hand over tasks or rely on the agent’s output.

Sense of Control- Collaborate over Delegate

  • Goal: The user wants to feel they can step in at any point—cancel or adjust tasks, provide additional input, or preview intermediate states.
  • Why: Transparency and interactivity reduce anxiety. Users remain confident if they can see what’s happening behind the scenes and know they can intervene. Moreover, at this time we want to ensure we can increase user satisfaction with outcomes. An approach of "Collaborative over Delegation" can support our ability to empower users to get the most of our Agentic offerings

Clarity and Progress

  • Goal: The user wants to understand the status of their requests: how far along the process is and what’s happening next.
  • Why: Clear progress indicators, partial outputs, and short updates assure users the system is actively working on their behalf.

FY26 Roadmap by Quarter

Quarter Key Initiatives

Q1 ( 1st use case in Private Beta)
  • Improve support for HiTL to allow for more collaboration
  • Enable Workflow to work with more than text inputs to include images
  • Enable remote execution to enable increased reliability and more compute intensive Workflows such as large scale refactoring

Q2
  • Enable internal teams to create specialized workflows, assisted by the Workflow team
  • System-level controls to enable customers to customize default context that WF needs to consider (Workflow policies)
  • Curate datasets for Agentic Workflows

Q3
  • System-level controls (Workflow policies)
  • Observability (workflow analytics, audit logs)
  • Workflow permissions
  • Enabling internal teams to create specialized workflows (Plan, Create, CI) self served

Q4
  • Enable customers to create and customize Agentic Workflows
  • Iteration based on user feedback

FY26 Detailed Items

Q1 (First Use Case in Private Beta)

Strategic Focus:

  • Improve “Human in the Loop” (HiTL) collaboration to build Trust & Confidence and a Sense of Control for early adopters.
  • Prepare Workflow to offer more platform capabilities

Key Initiatives:

  • Improve Support for Human-in-the-Loop (HiTL) Collaboration
    • What: Introduce real-time interactive checkpoints where users can review, refine, or override the Agent output.
    • Why (User Experience): Reinforces user Trust (system won’t run away with changes) and offers a Sense of Control (they can step in at will).
    • Outcome: Higher user satisfaction from collaborative Agent experiences, increased adoption among early testers.
  • Enable Workflow to Work with More than Text Inputs (Images)
    • What: Extend the Workflow engine to handle diverse content types (e.g., visuals for UI design feedback or code diagrams).
    • Why: Broader input types expand the scope of Agentic Workflows can support
      • Closer collaboration with more ways for the users to express themselves
        • Use case- Provide Workflow with an image of UI designs, this helps communicate to workflow the positioning and design guidelines that are relevant to a task
        • Use case- Provide Workflow with mermaid code diagrams to provide input to a code base structure it needs to implement when bootstrapping a new project or makes changes into an existing one
        • Use case- enable Workflow to to read data from images embedded in issues, epics, MR's etc.
        • The use cases above enable users to be able to collaborate better with Workflow, as they will be able to provide reach reference examples to help it achieve their desired outcome.
      • Use cases beyond coding—such design reviews, or documentation enhancements.
    • Outcome: Increased user ability to collaborate with Workflow.
  • Enable Remote Execution to Support Future Use Cases
    • What: Provide the ability to offload heavy tasks to external environments or cloud runners, while maintaining a clear link back to GitLab.
    • Why (User Experience): Increases GitLab ability to serve more use cases and more personas
    • Outcome: Scalability and performance improvements that make advanced, compute-heavy Agentic Workflow possible.

Q2

Strategic Focus:

  • Empower internal GitLab teams to create specialized workflows, expanding adoption across the platform.
  • Establish initial System-Level Controls to let organizations define default context and policies for AI usage, enhancing Trust & Confidence, and creating differentiator for our Agentic Workflow offering

Key Initiatives:

  • Enable Internal Teams to Create Specialized Workflows
    • What: Provide tools, templates, and guidance so GitLab teams (e.g., Plan, Create, CI/CD) can develop domain-specific Agentic Workflows.
    • Why: Demonstrates how Agentic solutions can be embedded throughout the SDLC, driving cross-functional adoption and showcasing GitLab’s innovation.
    • Outcome: Faster time-to-value for specialized user groups (Product Managers, DevOps, etc.), to support more use cases served by Agentic Workflows
  • System-Level Controls (Workflow Policies) to Customize Default Context
    • What: Allow enterprises to define default rules and contexts for Agentic Workflows—e.g., security constraints, compliance checks, coding standards.
    • Why: Ensures Sense of Control for organizations that need to maintain consistent guidelines across multiple teams.
    • Outcome: Foundation for robust governance, improving compliance and consistency for Agentic Workflow.
  • Curate Datasets for Agentic Workflows
    • What: Collect and refine domain-specific examples (historical MRs, code patterns, documentation) to evaluate Duo Workflow.
    • Why: High-quality data is critical for optimizing Workflow for specific use cases.
    • Outcome: Increased Workflow ability to serve more use cases, and to support mid-sized and enterprise customers at scale

Q3

Strategic Focus:

  • Deepen Observability & Compliance Controls (audit logs, permissions) to increase enterprise adoption.
  • Enable Self-Service for internal teams, ensuring they can build and manage specialized workflows independently, reinforcing the Clarity & Progress user pillar.

Key Initiatives:

  • System-Level Controls (Workflow Policies) – Expanded
    • What: Enhance policy options to include project-wide constraints, automated approval steps, and advanced configuration parameters.
    • Why: Organizations need Trust in consistent policy enforcement and a Sense of Control over how Workflow is applied across projects.
    • Outcome: Organizational controls will enable adoption with higher confidence.
  • Observability (Workflow Analytics & Audit Logs)
    • What: Introduce dashboards and logs for real-time insight into which workflows ran, what changes were proposed, and who approved them.
    • Why: To enable Agents at scale organizations to need to be able to audit agents action crossed with human employees to allow for accountability.
    • Outcome: Reduced risk, streamlined compliance audits, and provides ways to run Agents with human accountability.
  • Workflow Permissions
    • What: Role-based controls over creating, modifying, or executing Agentic Workflows.
    • Why: Aligns with organizational permission models, ensuring that only authorized users execute Workflows.
    • Outcome: Secure environment for AI-driven actions, minimizing accidental or unauthorized changes.
  • Enabling Internal Teams to Create Specialized Workflows – Self-Served
    • What: Provide documentation and tooling so that any GitLab team can independently build, deploy, and maintain their Agentic Workflow.
    • Why: Drives broader platform integration, fosters an internal ecosystem of AI capabilities, and accelerates time-to-market for new features.
    • Outcome: A more robust Agentic offering within GitLab that continuously innovates and meets varied user needs.

Q4

Strategic Focus:

  • Open up Agentic Workflows for external customers to build their own custom solutions, moving beyond GitLab’s internal usage.
  • Drive continuous Iteration Based on User Feedback, solidifying GitLab’s position as an industry leader in collaborative Agentic Workflows.

Key Initiatives:

  • GA Coding workflow
    • What: GA the first Workflow built on top of the workflow platform, Coding Workflow available to users via the IDE extension
    • Why: Establish GitLab market presence in the Agentic Market. Increase our costumers developer productivity
    • Outcome: Increase GitLab ARR with a new compelling product offering
  • Enable Customers to Create & Customize Agentic Workflows
    • What: Provide an external-facing Workflow Builder with templating, policy configuration, and data integration capabilities.
    • Why: Unlocks the ability for customers to design their own Agentic processes.
    • Outcome: Increases platform stickiness; organizations adopt GitLab as their Agentic Workflow solution for the entire SDLC.
  • Iteration Based on User Feedback
    • What: Continuously refine our offering based on feedback, usage analytics, and new market needs.
    • Why: Ensures we address real pain points and maintain a competitive edge.
    • Outcome: Highly satisfied user base, stronger renewal and upsell rates.
Edited by Pini Wietchner