Fostering User Trust in Agentic AI Tools: Insights from GitLab

Understanding Trust in Agentic Software

As artificial intelligence becomes more deeply embedded in modern software development workflows, user trust in AI-powered tools has become a critical success factor. At GitLab, we recognise that agentic tools—those that can perform tasks on a user’s behalf—must actively foster trust through thoughtful design choices that prioritise transparency, control, and safety.

The Importance of Transparency

From our user research, transparency surfaced repeatedly as a cornerstone of trust. Developers and DevOps teams want to know what an AI agent is doing, why it is taking an action, and what data it is relying upon. In response, GitLab has been working to ensure that AI-driven features provide clear context for their decisions and outputs. Whether it’s generating code or suggesting project configurations, visibility into the decision-making process is essential.

Consent and Control

Another theme echoed throughout our user interviews is the requirement for user consent before AI agents take action. Developers want to feel confident that they remain in control, especially when AI suggestions could impact production environments or sensitive data. GitLab’s approach emphasises confirmation prompts, granular permissions, and audit trails to uphold the user’s power of choice.

Trust Through Familiar Patterns

We found that users place more trust in AI tools when they behave in ways that are consistent with existing workflows and collaboration models. AI needs to augment—not replace—human capability. GitLab integrates agentic functions in ways that support collaboration across DevSecOps pipelines, reinforcing patterns users already know and trust.

Safety Nets for Confidence

Like any human assistant, an AI agent can make mistakes. Providing users with easy methods to undo, modify, or report erroneous actions is critical. GitLab implements version control and logs for every AI interaction, ensuring reliability and the ability to roll back unwanted changes.

Design Principles for Trustworthy Agentic Tools

• Explainability: Make it clear what the AI is doing and why.
• Agency: Give users decision-making authority over actions.
• Predictability: Ensure consistent, reliable behaviour.
• Safety: Design for recovery and non-destructive interaction.

Enabling Trust Across the DevSecOps Lifecycle

GitLab’s principal objective is to reduce cognitive load for developers and teams while keeping them empowered. Our agentic tools assist in planning, coding, CI/CD, and security testing—all while maintaining ethical AI standards. As adoption grows, we’re continuing to engage our community to learn and evolve our designs based on real needs.

Build Trust With GitLab Solutions

Are you preparing your organisation for the next generation of AI-powered development tools? IDEA GitLab Solutions offers expert GitLab consulting services, support, and licensing across the UK, Czech Republic, Slovakia, Croatia, Serbia, Slovenia, Macedonia, Israel, South Africa, and Paraguay. Let our certified professionals help you make the most of agentic tooling—securely, ethically, and efficiently.

Visit gitlab.solutions to learn more or get in touch with our regional experts.

• GitLab Docs Site Gets a Sleek New Design
• Protecting Customers with GitLab's Prompt Guardrails
• GitLab 18.7 Release: AI Code Suggestions, Compliance Reporting and More
• Boosting Higher Education IT Efficiency by Reducing Tool Sprawl
• Artois University Elevates Curriculum with GitLab Ultimate for Education