Greptile Reviews & Product Details

It actually understands the whole codebase, not only the line code changed. It actually makes a graph of my entire repo so it can tell after changes what affects my existing structure, so I can easily predict the consequences. It actually catches bugs, security issues, and also numbers the issues in a series. It also reads feedback and patterns we use, so it knows how we work, so it works better with time. It is already highly available and present in almost all IDEs and easily plugs into GitHub, GitLab, etc. Review collected by and hosted on G2.com.

I see one thing that while reviewing it takes a lot of time even if changes are smaller. I know it searches all code but it also needs optimization. Recent new pricing punishes us; it costs around 30 dollars per developer per month. If pull request volume increases, then it costs more than other competitors. They support through mail and Twitter, so there is no responsible person who listens to you, so your problem can take more time to resolve. It leaves more false comments which are not useful, but other things are good. Review collected by and hosted on G2.com.

The GitHub integration is seamless, once the repo is connected, Greptile automatically triggers a review on every pull request without any manual steps. The setup process is straightforward, and the dashboard gives a clean overview of PR activity, review scores, and bugs caught over time. The keyboard shortcuts in the Pull Requests tab are a thoughtful touch for power users. I also appreciate that it works silently in the background, fitting naturally into an existing dev workflow without disrupting it. Review collected by and hosted on G2.com.

All three of my test PRs, including one with a Python file containing a deliberate division-by-zero bug and a JavaScript file with undefined functions and missing null checks, were scored 4/5 with a "Low" severity rating. While the reviews completed quickly, the consistent scoring across very different code quality levels raises questions about how granular the risk assessment actually is. There's no severity breakdown per issue, and the dashboard doesn't surface what specific bugs were caught without clicking into each PR individually. A summary view of flagged issues across all PRs would make the analytics far more actionable. The persistent payment banner during the free trial also feels premature. Review collected by and hosted on G2.com.

What I like best about Greptile is how quickly it helps you understand a codebase without digging through files manually. You can ask direct questions and get clear answers tied to the actual code, which saves a lot of time when you’re trying to trace logic or figure out how something works. Review collected by and hosted on G2.com.

The only thing I dislike is that Greptile sometimes gives answers that feel a bit too high‑level when you’re trying to dig into very specific parts of the code. It’s great for general understanding, but occasionally you still have to jump into the files yourself to confirm the details. Review collected by and hosted on G2.com.

Unlike basic AI tools that only read the isolated PR diff, Greptile rebuilds a repository graph for each review. It analyzes how a change in one file might break dependencies, required parameters, or runtime logic in completely separate, seemingly unrelated areas of the codebase. Review collected by and hosted on G2.com.

Greptile functions strictly as a PR gate. Because it lacks native IDE integration for real-time coding feedback, developers only see its corrections after they have already committed to an implementation and opened a pull request. Review collected by and hosted on G2.com.

The full-codebase context is what really sets Greptile apart from a typical diff-only reviewer. Because it indexes the entire repository into a dependency graph, it can catch issues that ripple out from a change into files that weren’t even touched in the PR—exactly the kind of bug a surface-level reviewer tends to miss. On a few PRs, it flagged cases where a change to a shared utility would break something several files away, and we honestly hadn’t traced that impact ourselves during review.

The inline comments also come with clear explanations grounded in actual evidence from the codebase rather than generic suggestions. That made it much easier to trust what it flagged, instead of second-guessing whether it was just a false alarm. The automatic PR summaries save real review time too: getting a natural-language breakdown of what changed and why before diving into line-by-line comments makes the human review pass noticeably faster. Review collected by and hosted on G2.com.

Review turnaround is noticeably slower than lighter tools, since the deep multi-hop analysis across the whole codebase takes a few minutes rather than near-instant feedback — on a fast-moving PR cycle, that wait adds friction. The false-positive rate, while better than earlier versions, is still higher than we'd like, so some flagged issues need a quick manual dismissal rather than being fully actionable out of the box. Pricing is also worth watching closely — after the included review allotment, cost is per-review, so a busy sprint with a lot of PR activity can push the bill up faster than a flat subscription would. Review collected by and hosted on G2.com.

As an AI engineer managing a complex claims processing platform, reviewing pull requests across our backend monorepo is a massive time sink. I appreciate how Greptile acts as a true codebase-aware agent rather than just a standard auto-complete tool. Because it's integrated directly into our GitHub Actions workflow, it indexes the entire repository and provides highly contextual review comments whenever a PR is opened. I really value its cross-file understanding. For example, if a junior engineer modifies a core FastAPI Pydantic schema, Greptile successfully flags that this specific variable change will break our downstream LangChain JSON output parsers. Having an automated reviewer that actually understands the structural relationship between our API routing layer and our core machine learning extraction logic is invaluable. Review collected by and hosted on G2.com.

The initial repository indexing phase is a bit slow, especially as our codebase has grown to include massive infrastructure-as-code files and heavy data pipelines. Additionally, while Greptile is remarkably accurate with standard Python and FastAPI patterns, it occasionally hallucinates when suggesting hyper-specific PyTorch tensor shapes or referencing deprecated LangChain methods in PR comments. Because of this, senior engineering validation is still mandatory for complex machine learning diffs—it accelerates triage, but human oversight remains essential. On the enterprise side, deployment options and strict data governance controls could be streamlined, particularly for teams in heavily regulated sectors like insurance handling sensitive backend logic. Review collected by and hosted on G2.com.

What I like best is that Greptile actually understands the full codebase, not just the diff. It builds a graph of how files, functions, and dependencies connect, so it catches cross-file bugs and regressions that other reviewers—and honestly even human reviewers—tend to miss. The natural language search is also a game changer for onboarding: instead of asking a senior dev, “Where is this logic handled?”, you can just ask and get a grounded answer with file paths. It’s also learned our team’s review style over time, so the suggestions feel more relevant with each PR. It genuinely feels like having a senior engineer glance over every PR without slowing down the merge cycle. Review collected by and hosted on G2.com.

What I dislike is the noise-to-signal ratio on larger PRs. Greptile does catch real bugs, but it also flags a lot of nitpicks and false positives. Sometimes it leaves 8–10 comments on a PR where only 2–3 are actually worth acting on, and that ends up training developers to skim the feedback or ignore the bot entirely. And it Only supports GitHub and GitLab, so if your team is on Bitbucket or Azure DevOps, it's not an option at all. Review collected by and hosted on G2.com.

The GitHub integration with our repos and codebase gives it a wide array of context, instead of only looking at the context of pull requests. This is a big help when working on shared projects, because we don’t always have full visibility into how changes in other files might break dependencies. Greptile thinks through all of that in advance. The comments it leaves in PRs are also relevant, and the solutions it suggests are pretty sound. Overall, our whole SDK integrates very well with this. Review collected by and hosted on G2.com.

The prompt needs to be precise when working on a large project; otherwise, the generated output often isn’t very accurate. Filtering and narrowing down the issues can also be tough at times. Review collected by and hosted on G2.com.

What I like most about Greptile is that it checks the full codebase instead of only PRs. This is really useful when working on large projects, where a change in some files can easily affect or break other modules. The UI is great and easy to follow, so it’s simple to understand the overall flow. The GitHub integration is also solid, especially for including it in the PR workflow, and the review comments are great detailed and helpful. Review collected by and hosted on G2.com.

I’ve noticed that deeper codebase analysis can sometimes take longer when it’s a large repo or when PRs are bigger. Also, some of the comments aren’t very actionable and require manual inference before I can make changes. I’d like more control over filtering issues, so the most important ones are easier to identify. Review collected by and hosted on G2.com.

We have been using greptile for a long time and we like multiple things. 1. The comments on PR's are quite detailed and helpful with a button to fix directly with Claude, which gives a very good user experience. 2. Has very good integration with GITHUB. 3. Even on big PR's it gives the comments after analyzing in a very short time. Review collected by and hosted on G2.com.

1. Pricing could be better than their competitor. 2. Misses the overall picture of the PR - which could be improved though. Review collected by and hosted on G2.com.