Every engineering team has its own Terraform conventions — module structures, naming patterns, tagging strategies, remote state configuration. Off-the-shelf IaC generators ignore these conventions entirely, producing code that requires significant rework before it can be merged. DevOps Genie takes a different approach: it learns your conventions before writing a single line of code.
The learning loop works in three layers. The first layer is static analysis — when you connect your repository, DevOps Genie parses your existing Terraform modules to extract structural patterns: how you organize resources, which providers you use, how you handle variable defaults and output structures. The second layer is pull request feedback — every PR created by the AI and reviewed by your team generates a feedback signal. Approval patterns, requested changes, and inline comments all feed back into the model's understanding of what your team considers idiomatic.
The third layer is explicit convention configuration — you can define team-wide rules in plain English through the DevOps Genie dashboard. Rules like 'always use the company tagging module for resource tags' or 'S3 buckets must use the internal encryption wrapper module' become hard constraints the AI enforces on every generated PR. The result is that each Terraform PR produced by DevOps Genie gets progressively closer to what your team would have written themselves — without you having to define every convention upfront.