GitHub Copilot: Changing the Narrative from Productivity to Strong DevOps Foundations

In this post, Rob Bos challenges the current narrative around GitHub Copilot, advocating for a DevOps-focused approach rather than just productivity. He offers practical advice on integrating Copilot to strengthen engineering foundations and team processes.

GitHub Copilot - Change the Narrative

Date posted: 01 Apr 2025
Estimated reading time: 4 minutes
Author: Rob Bos

TL;DR

We need to shift the GitHub Copilot conversation from focusing solely on engineers and productivity, to emphasizing the importance of a strong DevOps foundation for accelerated and reliable delivery.
The next challenge is extending this approach throughout the organization.

Premise: The Current Narrative Isn’t Helping

Rob Bos argues that the existing narrative around GitHub Copilot is too narrow, primarily revolving around boosting developer productivity and simplifying code creation. This view, he explains, fuels the myth of ‘10x engineers’ and encourages risky practices such as ‘vibe coding’—rapidly accepting all AI-generated suggestions without careful validation or testing.

Such approaches, according to Bos, often lead to disappointment: incomplete or buggy code, AI making unnecessary changes that introduce bugs, and insufficient testing before production pushes. This focus can cause frustration and failed expectations, both for organizations and engineers alike.

He also cautions against over-emphasizing vague productivity metrics when evaluating Copilot’s impact, noting that productivity is hard to define and not always the best lens for measuring value.

A Better Narrative: Building a Sturdy DevOps Foundation

Rather than fixating on productivity boosts, Bos recommends harnessing Copilot and generative AI as enablers for robust engineering fundamentals:

Automated pipelines and testing
Everything as code or configuration
Peer review processes (“more eyes” principle)
Comprehensive testing to build trust
Continuous monitoring and feedback loops

Bos notes that once these basics are institutionalized, teams can deploy changes faster and with greater confidence, leveraging various types of testing (unit, regression, integration) suited to their contexts. He underscores that Copilot can accelerate not just code-writing, but also scripting, pipeline creation, and even writing queries, thereby freeing up time to tackle technical debt and backlog items.

Teams seeing the most success routinely dedicate time (e.g., 10% of each sprint) to technical debt reduction, facilitated by the speed-ups Copilot provides. With the right foundation, teams deliver greater value faster and enjoy fewer production issues.

The Next Level: Focusing on Team and Organizational Flow

Bos highlights that most engineers spend less than two hours a day on deep coding, with the ret of their time allocated to meetings, discussions, architectural planning, and documentation. He suggests that organizations should also empower roles like product owners to provide well-scoped, clear work descriptions—enabling engineers to add value more efficiently, especially as AI tools evolve.

Future advancements, such as features previewed in GitHub Copilot’s ‘project Padawan,’ could automate code change suggestions, testing, and even pull request creation. This elevates engineers to orchestrators or reviewers, focusing on high-trust oversight and broader system improvement.

Conclusion

Bos concludes that there will always be a place for skilled engineers, though their role may evolve. The focus should be on building trust in the AI-augmented systems, robust onboarding and training for new engineers, and adapting processes to help teams and organizations fully leverage Copilot and other AI tools. This cultural and practical shift enables more efficient work and ultimately better outcomes for end users.