Greptile AI
Greptile AI is an Artificial Intelligence tool designed to grasp the intricacies of any codebase, providing critical insights and support f...
Last verified:
What is Greptile AI?
Greptile AI — Greptile is an AI code review agent that automatically reviews every pull request with complete understanding of your entire codebase. Unlike traditional linters that check files in isolation, Greptile constructs a graph index of your codebase, building a detailed codegraph of how each function, class, file, and directory is connected. It then uses a swarm of agents to review changes, assess their impact beyond the diff, and flag potential issues that humans might miss, including style violations, security risks, and multi-file logical bugs.
Key features include full codebase context for reviews, conversational follow-ups where you can reply with @greptileai to ask fixing suggestions, reinforcement learning from feedback when you react with 👍 or 👎 to its comments, and pattern repos via greptile.json for repo-specific context. Greptile also offers TREX, an agent that writes and runs tests for every PR in a sandbox to catch bugs and missed edge cases. It integrates seamlessly with GitHub and GitLab, supports fix-with-your-agent buttons for Claude Code, Cursor, Codex, Conductor, and Devin, and offers Greptile MCP to connect with any AI agent.
Greptile is designed for software teams and developers working with large codebases, including enterprises across defense, healthcare, and financial services. Over 9,000 teams use Greptile, including companies like Stripe, Amazon, NVIDIA, Meta PyTorch, and Solana. It supports Python, JavaScript, TypeScript, Go, Elixir, Java, C, C++, C#, Swift, PHP, and Rust, with cloud or self-hosted deployment options including Docker, Kubernetes, and air-gapped environments.
The tool adapts to your team through custom rules written in plain English, continuous learning by reading your team's PR comments, and repository-specific context via patternsRepo. After 2-3 weeks, Greptile stops commenting on things you don't care about, reducing noise over time while catching 3X more bugs and enabling teams to merge 4X faster.
Greptile AI pricing
Pricing model: Paid
$30/seat/month. 50 reviews included per seat. $1 per additional review. Greptile offers free usage for qualified open source projects and significant discounts for pre-Series A startups. Contact them to check eligibility. Discounted bulk pricing for API use is available - contact [email protected] for more information. Enterprise plan includes self-hosted deployment in your own cloud with custom LLMs.
Greptile AI pros
- Catches bugs humans miss with full codebase context
- Reviews PRs in approximately 3 minutes average
- Learns your team's coding standards from PR comments
- Reduces noise over time after 2-3 weeks of use
- Conversational follow-ups with @greptileai replies
- Reinforcement learning from 👍/👎 reactions
- TREX agent writes and runs tests in sandbox for every PR
- One-click fix integration with Claude Code, Cursor, Codex, Devin
- Supports both GitHub and GitLab seamlessly
- SOC 2 Type II compliant with annual external audits
- Self-hosted deployment option for air-gapped environments
- Custom rules written in plain English for team standards
- Pattern repos via greptile.json for cross-repo context
- Supports 12+ languages including Python, Go, Rust, C++
- 9X faster time to merge for software teams
- 100K+ bugs caught in production every month
- Greptile MCP connects to any AI agent for comment context
- JIRA/Linear integration for ticket descriptions with codebase links
- No context switching - fix issues directly in your IDE
- 5 minutes setup to first review
Greptile AI cons
- $1 per additional review beyond 50 included per seat
- $30/seat/month may be expensive for small teams
- Lower response quality for languages outside the 12 main ones
- Does not run on past PRs by default
- Requires GitHub PAT with repo scope for indexing
- Bulk API pricing requires contacting support
- Self-hosting requires email inquiry for access
- Open source free usage requires eligibility check
Frequently asked questions about Greptile AI
What is Greptile and what does it do?
Greptile is an AI code review agent that automatically reviews every pull request with complete understanding of your codebase. It constructs a graph index of your codebase, then uses a swarm of agents to catch potential issues that humans might miss, including style violations, security risks, and multi-file logical bugs. With full-codebase understanding, Greptile can review pull requests, generate descriptions for tickets, diagnose Sentry alerts and failing tests, or integrate with any tools via a simple REST API.
How does Greptile's codebase context work?
When you connect your repos to Greptile, it generates a detailed graph of what every function, variable, class, file, and directory does, and how they are all connected. While reviewing a diff, Greptile uses this graph to retrieve the code affected by the diff, its dependencies, related code, and similar code, ensuring it has the context required to evaluate code changes beyond files in isolation.
What programming languages does Greptile support?
Greptile fully supports Python, JavaScript, TypeScript, Go, Elixir, Java, C, C++, C#, Swift, PHP, and Rust. Most other languages are supported with slightly lower response quality. Greptile works with any programming language in your codebase, but the 12 listed languages have the best response quality.
Does Greptile work with GitLab?
Yes, Greptile integrates seamlessly with GitLab in addition to GitHub. It supports both GitHub Cloud & Enterprise Server and GitLab Cloud & Self-Managed platforms.
How does Greptile learn from my team?
Greptile learns from your feedback when you react with 👍 or 👎 to its comments. This allows it to only comment on the things your team cares about most. Greptile also learns your codebase and coding standards by reading your team's PR comments. After 2-3 weeks, it stops commenting on things you don't care about, reducing noise over time.
Can I customize Greptile's review rules?
Yes, you can write standards in plain English and point Greptile at repo-specific context to enforce the patterns your team actually cares about on every PR. In the greptile.json file, you can specify a patternsRepo field with repos related to the one being reviewed that might add helpful context, such as adding a frontend repo to a backend repo's greptile.json file.
What is TREX and how does it work?
TREX is a Greptile agent that writes and runs tests for every PR in a sandbox to catch bugs and missed edge cases. It autonomously tests every PR to ensure code changes don't break functionality, providing an additional layer of validation beyond the regular code review agents.
Is Greptile secure and enterprise-ready?
Yes, Greptile is built for enterprises across defense, healthcare, and financial services. It is SOC 2 Type II compliant with independent audits and reports available on request. Greptile offers self-hosted deployment in your own air-gapped environment, SSO/SAML, audit logs, and dedicated support. The cloud product does not store customer code, instead pulling snippets as needed from your code host.
Can I self-host Greptile?
Yes, you can self-host Greptile in your AWS environment and even use your own LLM providers for added flexibility. Self-hosting options include Docker Compose, Kubernetes with Helm, and air-gapped environments. Email [email protected] to get access to their security report, inquire about self-hosting, or if you have any questions.
How do I get started with Greptile?
Sign up at app.greptile.com/signup using email, Google, Github, or GitLab. Install the GitHub/GitLab app and select the repos you want Greptile to review. Greptile will now review all new pull requests automatically. Setup to first review takes about 5 minutes. Greptile doesn't run on past PRs by default, but you can trigger the bot by commenting @greptileai on the PR.