In part 3 of this series, I argue that when AI makes implementation cheap, judgment, verification, and architecture discipline become the scarce capabilities that distinguish high-performing teams.
In part 2 of this series, I argue that context and policy form the control plane for agentic delivery. Clear task framing, trustworthy documentation, and policy at the point of execution let teams scale AI speed without scaling chaos.
In part 1 of this series, I argue that AI made code generation cheap while shifting the real constraint to delivery. Pull requests become the unit of flow, exposing pressure on review, testing, merge speed, and how humans and agents share work.
GitHub Agentic Workflows let you define AI-powered automation in natural language instead of YAML, unlocking a new Continuous AI loop. These workflows allow you to describe intent and convert that intent into actionable tasks executed by AI agents.
Using a script in my open source repo, you can download enterprise Copilot usage data and visualize PR activity by humans and Copilot.
From punch cards to AI pair programming to autonomous agents - a look at the evolution of developer productivity and what Agentic Software Delivery means for your future.
Teams must learn to ask “Should Copilot do this?” before starting work. This post shows how to teach async-first thinking, delegate routine tasks to AI, and redesign workflows for parallel experimentation.
Distributing GitHub Copilot licenses isn’t enough. Leaders must build intentional enablement programs with structured training, continuous support, and cultural change to unlock AI’s full potential across the organization.
Learn eight practical principles for evolving your software delivery with AI-driven capabilities while maintaining human oversight and delivering real business value.
Turn flaky or failing builds into a self-healing loop with GitHub Actions, GitHub Models, and Copilot. Automatically analyze failures, classify root cause, and open a remediation issue when it is not transient.
