Tackling SLO cognitive load with AI — coach design, automate the rest

Nulab's platform engineering team tackled a common challenge: engineers knew their services but found SLO practices and Datadog and Terraform setup daunting, leaving configuration to SREs and eroding ownership. They split the cognitive load between what teams must understand and what to automate. For design, their AI agent coaches through dialogue, pulling metrics from Datadog's MCP Server so teams can make data-driven decisions without tool expertise while understanding their SLOs. Once complete, a Terraform module with best practices automates Datadog setup. Drift detection and review reminders keep SLOs current.

Takeaways:
• Splitting cognitive load: what to coach vs. what to automate
• Building a self-service SLO workflow with AI and Datadog MCP
• Preventing SLO staleness with automated drift detection