Spec-driven development: The AI engineering workflow at Notion | Ryan Nystrom

AI-driven workflows are fundamentally reshaping engineering management by shifting the focus from administrative overhead to high-level architecture and systems thinking. By deploying custom agents to synthesize daily context from Slack, GitHub, and telemetry, engineering teams can eliminate manual meeting preparation and foster more productive, data-informed stand-ups. This shift extends to the development process itself, where spec-driven development—using natural language to define features before invoking autonomous agents to write and verify code—accelerates delivery cycles. Maintaining a high-velocity CI/CD pipeline is critical to this model, as it provides the rapid feedback loops necessary for agents to function effectively. Ultimately, these tools allow engineering managers to remain hands-on with code, reducing burnout while increasing overall team output and innovation.