Valudio automates the first pass on every pull request in Azure DevOps with the AI code review tool CodeRabbit. The review cycle for a well-scoped pull request typically dropped from over an hour to about 15 minutes. Integration took three working days for a team that builds in .NET, Node, TypeScript and React. Most of that time went into Azure’s permission model, not the tool.
In conversation with Dennis Quitaquís Tamay, Tech Lead at Valudio. Leda Felix asked the questions.
“A pull request isn’t something you can easily plan for. It’s completely asynchronous: the moment a developer finishes something or addresses our feedback, it needs to be reviewed almost immediately, or it becomes a bottleneck before QA can even start testing.”
Dennis Quitaquís Tamay, Tech Lead, Valudio
A pull request is code a developer submits for a colleague to check. It arrives when the work is done, not when the reviewer has time. Three frontend and two backend developers were pushing pull requests faster than one person could read them, and the pile in front of QA grew. The bottleneck was a single step, the first look at every pull request, and it sat with Dennis.
Where the code review bottleneck came from
Every pull request went through two review phases. The first asked whether the code was clean and met our quality standards. The second asked whether what was built actually matched the user story. Backend and frontend were reviewed separately, and most backend pull requests landed with Dennis.
Depending on the quality of the first version, that meant zero, one or several rounds: feedback, correction, another review, until both sides agreed. Static analysis tools like SonarCloud and Snyk run alongside, but they cover different ground: risk and vulnerabilities, not design and logic.
The Tech Lead’s time disappeared into waiting and re-reading, and QA waited for him. Nobody was doing bad work. A step that has to happen right away was waiting for the person with the least time.
The workaround that did not work
Before any tool was involved, the team pasted code changes by hand into a general-purpose AI chat to have them sanity-checked. It helped a little. It was also one more manual step on top of a manual process, and it was never an answer to the problem.
Someone copies something into a chat window, gets something useful back, and the workflow stays as it was. That is what AI looks like in many companies. It feels like progress. It is not, as long as the bottleneck sits in the same place. Ours stayed with Dennis, chat window or not. We were left with one person who owned the problem, and one step that was too slow.
We cover why working groups with plenty of ideas stall at exactly this point in “Why AI working groups deliver nothing”.
What we needed to automate pull request reviews
Three criteria, nothing more. The review has to start automatically the moment a pull request opens, without anyone launching it. It has to run natively in Azure DevOps, the backbone of our Microsoft environment. And it has to work with everything we build in: .NET, Node, TypeScript and React.
Dennis ran the search with the team, so the person who received most of the pull requests chose the tool. Besides CodeRabbit, they looked at CodeAnt AI and Aikido:
| Tool | Why it fitted, or did not | Decision |
|---|---|---|
| CodeRabbit | Native integration with Azure DevOps, starts automatically when a pull request opens. Confirmed in a two-week trial with real pull requests in .NET, Node, TypeScript and React. Complements SonarCloud and Snyk rather than duplicating them. | Adopted |
| CodeAnt AI | Appears oriented toward Jira and GitLab, not Azure DevOps. Not tested in our environment. | Not pursued |
| Aikido | Not tested: the overlap with the security scans we already run through SonarCloud and Snyk settled it before we got that far. | Not pursued |
CodeRabbit was the first result in a search for an AI pull request tool for Azure. The two-week free trial settled it. From day one we tested it on real pull requests from our own projects, before any budget was on the table.
“Its integration is oriented mainly toward Microsoft, so every step of the setup involved things we already recognized and could execute ourselves. There wasn’t a single strange or unidentifiable step.”
Three days, and the hardest part was not the tool
“The full integration, I’d say, didn’t take more than three continuous days: first finding the solution, second doing the testing, and third putting it into practice and configuring it.”
One day per step. The hardest part was not the tool but Azure’s permission model, with users and subscriptions spread across many client tenants. The team first had to work out which user in which tenant and which subscription actually had access to the projects CodeRabbit should review. Once that was settled, the rest went quickly. CodeRabbit received read access to the code and nothing more, and since then it has commented on every new pull request automatically. The effort of adopting AI sits in your environment, not in the tool.
What changed for the team with AI code review
Today CodeRabbit takes the first pass on every pull request: it flags quality issues, suggests better variable names, recommends splitting a method in two, and covers the part that used to run entirely by hand. The suggestions stay suggestions. The developer decides whether to apply them.
“Now we go directly to understanding the code, directly to the architecture, directly to the requirements of the user story. We look less at the detail and more at the design the developer chose for the solution.”
What used to add up to over an hour, once every round was counted, now typically takes about 15 minutes for a well-scoped pull request. Complex cases still get a real conversation, about logic and hidden bugs, not formatting.
“The AI’s suggestions are entirely recommendable, but not obligatory, and that’s important for us. We can choose not to follow them. It can feel a bit overwhelming at first, seeing 10 comments appear on a single file you’ve changed. But those are 10 comments you don’t have to apply: you read them, interpret them, and then decide.”
The backlog in front of QA is gone, and QA no longer waits on a review to start testing.
What this means for you
Our bottleneck had a name, Dennis, and a step: the first look at every pull request. It did not need an AI roadmap. The tool fit the environment we already run, and it worked on real pull requests for two weeks before we talked about a budget. That is the pattern we follow on AI projects with clients: one owner, one concrete bottleneck, a trial with real work before a budget. The only difference here is that we were our own client. Where AI does not help in our daily work is the subject of “The downsides of AI in everyday development”.
Frequently asked questions
Does AI code review work with Azure DevOps?
Yes, if the tool plugs into Azure DevOps directly instead of going through another system. At Valudio, CodeRabbit comments on every pull request the moment a developer opens it, with read access to the code and nothing else. The tool was the easy part. Budget your time for Azure’s permission model instead. Which user, in which tenant, under which subscription, actually reaches the repositories you want reviewed? That question took most of our three integration days.
Does AI code review replace human reviewers?
No. It replaces the first read, the pass that used to catch naming, style and methods that had grown too long. At Valudio a Tech Lead or a colleague still approves every pull request. The difference is what they spend that approval on: the design the developer chose, and whether the result matches the user story. A well-scoped pull request now takes about 15 minutes instead of over an hour. Complex ones still get a conversation.
How long does it take to implement an AI code review tool?
Plan for days, not months, if the tool fits your environment. At Valudio it took about three continuous working days: one to find the right way to integrate, one to test, one to configure and roll out. The setup itself was routine. The time went into working out which user, tenant and subscription controlled the projects across our client tenants. Before that, a two-week free trial on real pull requests had settled the choice, so no budget decision was waiting on the outcome.
Can developers ignore the AI’s suggestions?
Yes, and at Valudio that is the rule, not the exception. CodeRabbit’s comments are recommendations, and the team treats them that way. A developer reads them, keeps what improves the code and drops the rest. The first pull request with ten comments on one file feels like a lot. After a few rounds, reading them becomes a habit, and the developer, not the tool, decides what ships.
Let’s talk about your bottleneck
If your team has a similar bottleneck, we are happy to talk it through. First call, 30 minutes. We will tell you honestly whether your plans and budget fit together, and what a discovery and concept phase would look like for you. We reply within 24 hours. Book a call.
