Datadog launches Pup CLI to bridge AI agents and observability data

The release of Pup CLI marks a shift toward 'agentic observability,' where AI is not just a dashboard assistant but an active participant in infrastructure management. For high-scale streaming environments, this reduces the Mean Time to Repair (MTTR) by allowing autonomous agents to correlate traces and remediate issues without human intervention. By standardizing the interface for terminal-native agents like Claude Code and GitHub Copilot, Datadog is positioning its API as the essential telemetry layer for the next generation of AI-driven DevOps. Watch for whether this reduces operational overhead as streaming providers struggle with the data deluge from complex multi-CDN and edge-heavy architectures.

Datadog's Pup CLI arrives as the programmatic complement to its Bits AI agent suite, which the company has been rapidly expanding since mid-2025. Datadog launched Bits AI SRE at its DASH conference in December 2025 as its first generally available AI agent, designed to investigate alerts and surface actionable root cause in minutes. The agent had already been tested against more than 2,000 customer environments with tens of thousands of investigations run, according to Chief Product Officer Yanbing Li. Pup CLI's structured JSON output and OAuth-scoped authentication extend this agentic philosophy to terminal-native workflows, giving external AI agents the same telemetry access that Bits AI SRE uses internally.

By March 2026, Datadog had significantly upgraded the underlying agent architecture. The company announced that Bits Investigation now completes investigations approximately twice as fast, in roughly 3-4 minutes, with tighter integration to MCP-powered tools. The update introduced direct triage actions from within the chatbot experience, including creating incidents, paging engineers, and generating Jira tickets, plus three new actions in the Datadog Action Catalog for automated remediation. Pup CLI's self-discoverable commands and 33-domain coverage position it as the interface layer that third-party agent frameworks can use to invoke these same capabilities without building custom API wrappers.

Datadog has also published technical detail on how its agent architecture handles the complexity of distributed systems. The company reported in January 2026 that Bits Investigation is already helping teams decrease time to resolution by up to 95%, a figure that underscores the operational stakes for streaming providers running multi-service architectures. The agent reasons over telemetry autonomously, exploring exponentially more hypotheses in parallel than a human engineer could, which is the same design philosophy behind Pup CLI's machine-readable output format. For streaming infrastructure teams evaluating observability stacks that support autonomous remediation, Pup CLI's Rust-based performance and broad product coverage represent a concrete differentiator in the emerging agentic observability category.