2025 raised the bar for AI code reviews at Bito.
We spent the year focused on one goal, to build a smarter and more helpful Code Review Agent for real engineering teams.
Across Git platforms, IDEs, and cloud features, the agent grew in ways that shaped how teams write and review code every day. This post brings together every major update we shipped in 2025 and shows how the product moved forward across the year.
Let’s get into it.
Agentic review engine
The agentic review engine was a major step in how the Bito’s AI Code Review Agent understands your code. It builds its own context, reads related files, and confirms issues with real evidence before posting feedback.
Our agentic review engine delivers evidence-based feedback. Here’s how:
- Agent uses deep reasoning and code analysis to determine if an issue exists.
- Bito’s system intelligence layer can span services and cross-repo components to deliver entire codebase context.
- An evidence-based approach confirms each issue and reduces noisy, low value suggestions. Evaluations show a 57% reduction in false positives.
- Bito’s learning system adapts to your team’s historical reviews and updates its understanding of what really matters to your org.
This upgrade shaped many of the improvements that followed in 2025 and became the core of how the agent reviews code today.
AI code reviews in your IDE
This was the biggest update of the year. We brought AI code reviews in your IDE like VS Code, JetBrains, Cursor, and Windsurf.
Bito is changing how teams review code by enabling the agent to work on your local files before any pull request. It reads your changes, builds context, and points out issues while the work is still in front of you.
You can review your work in several ways inside the IDE:
- review all local changes
- review staged changes
- review any file or folder
- target a single commit or a range of commits
Trigger these reviews through the @codereview command in the IDE chat, the right click menu, or the command palette. Here’s a walkthrough:
Cross repo and cross service context
Bito’s AI Code Review Agent now pulls information from related files, follows logic across services, and checks connected parts of the project before it posts a suggestion.
This grew out of the agentic review engine and our improvements to symbol search in large JS and TS projects.
With these upgrades, the agent handles exports, variables, and control flow with more accuracy. This helps reviewers see real issues that come from the wider project, not only the lines shown in the diff.
If you work with multi service repos, you can read more here:
Project aware reviews
Bito now reads and uses guideline files that are commonly used by AI coding agents like Cursor, Windsurf, and Claude Code.
Simply add guideline files (.cursor/rules/.mdc, .windsurf/rules/.md, CLAUDE.md, GEMINI.md, or AGENTS.md) to your repository, and the agent automatically applies your project’s standards when reviewing pull requests.
Repo level controls with .bito.yaml
Repo-level Agent settings let you control how the AI Code Review Agent behaves for each repository. By placing a .bito.yaml file in the root of your repository, you can define custom review preferences that apply only to that repository.
Bito automatically detects the presence of a .bito.yaml file in a repository and applies its configuration to override the global Agent settings defined by admins in the Bito Cloud UI.
Platform integrations and workflow support
i/ Jira integration
The Code Review Agent can now validate pull requests against Jira ticket requirements. When a pull request references a Jira ticket, the agent reads the linked issues, extracts the requirements, and checks the code against each point.
The results appear in a table that marks items as met, missed, or partial. This feature supports Jira Cloud and Jira Data Center and can be enabled per agent in the integration settings.
Here’s a video demo:
ii/ Perforce and SVN support
The IDE extension works with Perforce and SVN. Developers can run reviews on these codebases in the same way they review Git projects. The agent reads local changes, builds context, and shows suggestions inside the IDE panel.
iii/ Multi group support for GitLab self-managed
Bito supports multiple GitLab groups for GitLab self-managed installations. You can add or remove groups from the Repositories page and switch between single or multiple groups as needed.
Here’s the guide for GitLab for GitLab (Self-Managed)
Learned rules and custom guidelines in Bito
Bito can follow your team’s style through learned rules and custom guidelines.
- Learned rules come from the feedback you give on Bito suggestions. When a pattern receives enough negative signals, the agent turns that rule on and avoids similar comments in future reviews.
Now this is one of the most requested features of Bito: Custom guidelines. These let you define the rules the agent should follow.
- You can create these rules in the Bito dashboard for a specific language or for all languages, then apply them to any agent in your workspace. When a guideline triggers a comment, the agent shows a citation that points to the exact file and section, so developers know where the rule came from and why it was applied.
Here’s a video walkthrough:
Review modes and multilingual support
You can switch between Essential and Comprehensive review modes based on how much detail you want in code reviews. Essential keeps the focus on important issues, while Comprehensive adds smaller points that help with cleanup and consistency.
The agent also supports more than twenty languages, including English, Hindi, Chinese, and Spanish. This helps teams review code in the language they are most comfortable with and keeps feedback clear across regions.
… and a lot more! Go through Bito’s changelog to see all the updates we’ve shipped:
Request Changes and cleaner PR feedback
Bito supports Request Changes comments across GitHub, GitLab, and Bitbucket. When this setting is on, the agent posts important issues as formal change requests that follow your branch protection rules.
If your workflow requires all threads to be resolved before merging, these comments block the pull request until the issue is fixed. This prevents merges that skip important fixes and helps teams keep quality steady as the codebase grows.
Here’s a video walkthrough:
Chat with Bito in your PR
This is one of our most interesting updates of 2025. You can now talk to Bito inside a pull request.
When you reply to one of its comments or tag @bitoagent or @askbito, the agent can explain why it flagged a line, offer another way to fix the issue, or show an example that makes the change clear.
It supports more than 20+ human languages and responds within seconds. You can read about its real-life use case here: Chatting with My AI Code Review Agent
Here’s a quick demo of the feature:
PR and issue level analytics
The PR Analytics dashboard gives teams a clear view of how reviews perform across the codebase. You can see which pull requests contain issues, which ones passed clean, and which were skipped due to configuration limits.
Each pull request view shows issue types, categories, and links to exact lines in the code. This helps teams track patterns, understand frequent problem areas, and follow code quality trends across the project.
Here’s a quick walkthrough:
Bito in 2025 by the numbers
Bito reviewed code at a scale we have never seen before. The agent ran across hundreds of teams, touched millions of lines, and flagged issues that helped developers ship cleaner code.
- 350K+ pull requests reviewed
- 11K+ developers reviewed code with Bito
- 130M+ lines of code reviewed
- 7K+ repositories connected
- 400K+ issues found and surfaced in reviews
- 53% reduction in bugs
- 40% suggestion acceptance rate
Here is what teams shared about their experience:
“Bito gives targeted, context-aware feedback that reduces the load on reviewers and improves overall quality. It is a valuable part of our daily development workflow.”
– Kumar Vishnu, Director of Engineering at Apica
“Bito is like a smart pair of eyes that watches over the code we are shipping. It catches small human mistakes that can affect an entire pipeline.”
– Charles Guerin, CEO at OBDS
by the Bito team. 
