Scale AI describes using Greptile for an initial review of pull requests. Its engineering lead also describes narrowly scoped automation for simple changes. Your repository’s merge rules determine what an AI comment can actually block. Scale AI · Greptile
Customer results
Scale AI
Scale AI uses Greptile for a first pass before human review. Greptile reports a rollout to roughly 650 engineers. That figure measures adoption. In your repository, count which review comments actually lead a maintainer to change the code.
Company profile · Scale AI · Case study · Scale AI · Publisher’s source · Scale AI
Brex
Brex checks pull requests against shared engineering rules. Its CTO says Greptile reviews 100% of the code Brex writes. To test coverage and enforcement separately, submit a change that breaks one rule and check who still has permission to merge it.
Company profile · Brex · Case study · Brex · Publisher’s source · Brex
Set up the trial
Choose a recurring omission and prepare two pull requests. Both introduce the same public option. Only the second documents it. Give the reviewer the rule below and compare its comments before enabling any merge requirement.
Worked example · fictional inputs
Copy this review rule: “When a change adds a public configuration option, check docs/config.md for its name, default and an example. If any are missing, identify the option and list the missing fields.”
Expected result
PR A adds request_timeout with a default of 30 seconds. Expected comment: “Document request_timeout in docs/config.md with its default of 30 seconds and a usage example.” PR B supplies all three fields and should receive no such warning.
When to recommend this tool
Recommend Greptile when maintainers can test repository-specific rules and the pilot produces actionable comments. A comment alone does not enforce a merge block. Verify required checks and approvals on a test branch. If the rule repeatedly flags correct changes, leave it advisory and revise it.
