AI Does Not Make Operating IaC Cheap

Hey folks,

Something that is continually proven when working with clients here at Masterpoint: AI made writing IaC cheap.

But, it did not make operating IaC cheap.

Because the problem was never “can we write the code faster?”

The problem is everything that happens after the code is written:

1. How does code move from “pushed to PR” => “validated” => “security checked” => “merged” => "applied to production".

2. What do you enforce so every repository and/or team doesn’t invent their own snowflake workflow?

3. How do you avoid the rough edges (including the ones that only show up at scale).

4. How do you manage new root modules and workspaces sustainably.

5. Who approves what? And how can we make sure this is auditable?

6. Who gets notified when an apply fails? How fast can they recover?

If your “IaC speed” problem feels unsolved, there’s a chance you don’t have a code problem.

You have got a workflow/pattern/ownership problem. And more AI written code can’t help with that.

I'm curious what the bottleneck is in your org is right now:

• Writing the IaC?

• Or getting it safely to prod?

Hit reply and share your thoughts!

Matt @ Masterpoint

PS How time flies! Here’s a throwback video of when I joined Adventures in Devops in March 2025 to talk automation strategies.