Datadog Work Management: One Queue for Humans and AI Agents
Enterprise Developer PlatformsMost engineering organizations are now running a mix of human and AI agents on operational work. The tracking problem is worse than it looks: when an incident fires, it may be investigated by a human, triaged by an agent, escalated back to a human for approval, and resolved by another agent — across three different systems with no shared record of what happened and why.
What Happened
Datadog relaunched its Case Management product as Work Management, reframing it as the coordination layer for both human and AI agent work rather than just a ticketing system attached to monitoring.
The core change is conceptual as much as functional: each work item in Work Management retains a live connection to the Datadog signals that originated it — logs, traces, monitors, service ownership records. When a case is closed or escalated, that context travels with it. Slack discussions are automatically captured in the activity timeline. Human approvals are gate-enforced before AI agents proceed on sensitive actions.
Integration breadth is bidirectional: Jira, Linear, ServiceNow, and GitHub sync in both directions, so Work Management can function as an orchestration layer rather than requiring teams to migrate away from existing tools.
Why It Matters
The fragmentation problem Work Management targets is real and specific. When work is tracked in spreadsheets, Slack threads, Jira tickets, and alerting tools simultaneously, the cost is not just confusion — it’s lost context. An incident that was diagnosed, mitigated, and reopened three times in six months doesn’t show that history when the fourth recurrence hits. An AI agent handed a task with no knowledge of what was tried before will repeat the same failed approach.
Work Management addresses this by making the history of a work item observable, not just the current state. The MTTA (Mean Time to Assign) and MTTC (Mean Time to Close) metrics are tracked in-platform, creating a feedback loop between how work is organized and how long it takes to resolve.
The AI coordination layer is the part that distinguishes this from a Jira refresh. Agents built with Datadog’s Bits Agent Builder can be assigned to work items, perform investigations and code changes, and escalate to human reviewers when they need clarification — with a hard requirement for human approval before any critical action executes. The agents operate within the same observability context as human engineers, with access to the same logs and traces.
The So What
Work Management does not solve the underlying complexity of coordinating human and AI work. What it does is make that coordination visible, with provenance attached. For platform and engineering teams running mixed human/agent operations, the practical question is whether your current tracking approach gives you a complete history of any given incident or task — including which actions were taken by agents, which were approved by humans, and what was tried and abandoned. If the answer involves reconstructing that history from Slack and Jira and alerting dashboards, Work Management’s single-context approach is worth evaluating. The audit trail requirement is already here; the tooling is catching up.
Source: Datadog Blog
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →