AI in Infrastructure as Code: Kief Morris on Navigating the Limits of Automation

Josh Pollara • July 9, 2025

Kief Morris

Distinguished Engineer at Thoughtworks

Infrastructure as Code Pioneer & Author

About Kief

Kief Morris is a Distinguished Engineer at Thoughtworks and the author of the seminal book "Infrastructure as Code." With over two decades of experience in software delivery and cloud infrastructure, Kief has been at the forefront of defining and evolving infrastructure automation practices. He helps organizations worldwide adopt modern infrastructure management approaches, focusing on principles of consistency, repeatability, and governance that make infrastructure reliable and manageable at scale.

Current Focus: Measuring Infrastructure Automation Effectiveness

Kief is currently developing a series of articles exploring how to measure and improve the effectiveness of infrastructure automation. His approach focuses on understanding the user journeys and value streams of the people who use the infrastructure, particularly software development teams. Through his blog and email newsletter, Kief shares these evolving insights with the infrastructure community.

The key is balancing GenAI's speed with ensuring infrastructure remains consistent, reliable, and governed.

Q1: Where have you seen AI genuinely transform infrastructure, and where's it mostly hype?

I've mainly leveraged GenAI to build tools that help manage infrastructure codebases, significantly aiding my learning curve in GoLang. Currently, I find GenAI most effective as a learning assistant rather than a direct code generator for infrastructure. It doesn't yet understand infrastructure code as deeply as general programming languages, although this is rapidly improving. I'm particularly concerned that the infrastructure code used to train large language models isn't always well-designed, reflecting the industry's ongoing journey to establish best practices.

AI-Assisted ClickOps might overlook fundamental cloud infrastructure principles, such as consistency across environments, repeatability, and transparency.

I've also observed "AI-Assisted ClickOps," where GenAI provisions infrastructure from simple prompts. While compelling, I worry this approach might overlook fundamental cloud infrastructure principles, such as consistency across environments, repeatability, transparency, and ease of governance and troubleshooting.

Q2: How are team roles evolving as AI becomes commonplace?

Just as before GenAI, it remains crucial to have specialists who create tools, components, prompts, tests, and resources, enabling software teams to independently provision the infrastructure they need. The critical evolution lies in building platforms and tools that allow specialists like security engineers, platform engineers, and others, to contribute their expertise in empowering ways.

It's about embedding expert knowledge directly into tools and workflows, ensuring teams have self-service capabilities without compromising security or quality.

AI enhances this model by helping teams swiftly acquire the infrastructure they need, provided the platforms and toolsets are designed thoughtfully.

Q3: Fast-forward five years: how do you see us managing infrastructure day-to-day?

In five years, GenAI will play a role in guiding software teams to clearly define infrastructure needs. For example, AI could ask detailed questions about application requirements like inbound connection rules or data sensitivity to build precise infrastructure specifications. These specifications must remain persistent to ensure deterministic, consistent, and transparent infrastructure provisioning, regardless of deployment environment. Additionally, GenAI will empower platform domain and governance specialists by aiding them in creating and validating infrastructure components that other teams rely on. Developer portals enhanced with chat interfaces can quickly create specific environments such as performance testing setups, requested by consumer teams.

The critical aspect will be balancing GenAI's ability to rapidly fulfill infrastructure requests with the need for consistency, reliability, and rigorous governance.

Josh's Note

Kief’s take on balancing AI speed with staying consistent and transparent hits home. At Stategraph, we see the future of infrastructure as inherently collaborative, expert-driven platforms enhanced by AI, but always grounded in clarity and governance.

Follow Kief's work:

Infrastructure as Code Blog - Deep insights on infrastructure patterns and practices
LinkedIn - Connect for updates on infrastructure automation and governance
Infrastructure as Code Book - The foundational guide to modern infrastructure management

More in the Series

Justin O'Connor on Accelerating the Right Way →

Petar Nikov on Avoiding AI Pitfalls →

Get Started Read the Docs