A familiar situation
The agents are waiting. Nobody's using them.
GitLab Duo Agent Platform became generally available in January 2026. For organisations on Premium or Ultimate, that means AI agents across the full software development lifecycle are already included in their credits — planning, coding, security, CI/CD, and more.
And yet for most organisations, adoption is near zero.
The reasons are consistent. The instance isn't on a version that supports Agent Platform. The AI gateway isn't configured correctly. Engineers have been told the feature exists but haven't been shown what it does in their day-to-day workflow. Nobody has taken ownership of driving adoption. The credits are there. The agents are waiting. And the seven hours a week lost to manual handoffs, context switching, and review backlogs continues uninterrupted.
The organisations getting the most from GitLab Duo Agent Platform aren't the ones with the most advanced engineers — they're the ones that treated enablement as a deliberate programme, not an announcement.
The pattern
AI agents across the full development lifecycle are already in the licence. Configuration and enablement are what's missing.
AI speeds up coding. Fragmented tools and manual handoffs still cost teams an average of seven hours per person per week.