As monday.com’s engineering organization expanded beyond 500 developers, the company faced a significant challenge: the rapid increase in the volume of code being produced. With product lines multiplying and microservices proliferating, the sheer number of pull requests (PRs) submitted each month began to overwhelm the human reviewers tasked with ensuring code quality. The need for a solution that could efficiently manage this influx while maintaining high standards of software quality became paramount.
In response to this challenge, Guy Regev, the Vice President of Research and Development at monday.com, initiated an exploration into AI-driven solutions. This led to the adoption of Qodo, an Israeli startup specializing in developer agents. Unlike traditional code generation tools such as GitHub Copilot or Cursor, which focus on writing new code, Qodo is designed specifically for code review. It employs a novel approach known as “context engineering,” which allows it to understand not only the changes made in a pull request but also the underlying reasons for those changes, how they align with business logic, and whether they adhere to internal best practices.
The implementation of Qodo marked a transformative shift for monday.com. Regev described the tool as akin to adding a new developer to the team—one that learns the intricacies of the organization’s workflows and coding standards. Since its integration, Qodo has successfully prevented over 800 issues per month from reaching production, including potential security vulnerabilities that could have had serious implications for the company.
At the core of Qodo’s functionality is its ability to conduct code reviews at scale. As monday.com’s developers work across hundreds of repositories and services, the need for a system that can evaluate pull requests against team-specific conventions, architectural guidelines, and historical patterns became evident. Qodo addresses this by learning from the organization’s own codebase, training on previous PRs, comments, merges, and even Slack discussions to gain insights into how the team operates.
The concept of “context engineering” is central to Qodo’s effectiveness. This approach involves a comprehensive analysis of all relevant information when making decisions about code quality. It considers not just the code diff presented in a pull request but also prior discussions, documentation, relevant files from the repository, test results, and configuration data. By leveraging this extensive context, Qodo can identify not only obvious bugs but also subtle issues that might escape human reviewers, such as hardcoded variables, missing fallbacks, or violations of cross-team architectural conventions.
One notable instance highlighted by Regev involved a recent pull request where Qodo flagged a line of code that inadvertently exposed a staging environment variable. This critical oversight went unnoticed by human reviewers, but had it been merged, it could have led to significant problems in production. The potential hours spent rectifying such a security leak—and the legal ramifications that could follow—far outweigh the time saved during the initial review process.
Today, Qodo is deeply embedded within monday.com’s development workflow. It analyzes pull requests and provides context-aware recommendations based on prior team code reviews. This integration has been seamless, as Qodo operates directly within GitHub through pull request actions and comments, eliminating the need for developers to learn a separate tool. Instead, it feels like an additional teammate who understands the team’s unique dynamics and coding standards.
The human-in-the-loop model employed by Qodo ensures that developers retain control over final decisions during the review process. They receive suggestions and feedback from Qodo, which helps them learn from their peers and take ownership of their code. This collaborative approach fosters a culture of continuous improvement and knowledge sharing among team members.
Since the broader rollout of Qodo, monday.com has experienced measurable improvements across various teams. Internal analyses indicate that developers save approximately one hour per pull request on average. Given the thousands of PRs submitted each month, these time savings quickly accumulate, translating into thousands of developer hours saved annually. Importantly, the issues flagged by Qodo are not merely cosmetic; many pertain to critical aspects such as business logic, security, and runtime stability. Because Qodo’s suggestions are tailored to reflect monday.com’s actual conventions, developers are more inclined to act on them.
The accuracy of Qodo’s recommendations stems from its data-first design. By training on each company’s private codebase and historical data, Qodo adapts to different team styles and practices, avoiding the pitfalls of one-size-fits-all rules or external datasets. This level of customization ensures that the tool remains relevant and effective in addressing the specific needs of monday.com’s engineering teams.
Regev’s team was so impressed with the impact of Qodo that they began planning deeper integrations between Qodo and Monday Dev, the developer-focused product line that monday.com is developing. The vision for this integration is to create a workflow where business context—such as tasks, tickets, and customer feedback—flows directly into the code review layer. This would enable reviewers to assess not only whether the code functions correctly but also whether it effectively addresses the right problems.
Before adopting Qodo, monday.com relied on traditional tools such as linters, danger rules, and static analysis. These rule-based systems required extensive configuration and often failed to account for nuances that only experienced engineers would recognize. In contrast, Qodo’s approach feels more intuitive, as it learns from the engineers themselves, adapting to their workflows and preferences.
Looking ahead, Qodo is expanding its platform to include additional tools designed to enhance the developer experience further. Among these are Qodo Gen, which focuses on context-aware code generation; Qodo Merge, which automates PR analysis; and Qodo Cover, a regression-testing agent that utilizes runtime validation to ensure comprehensive test coverage. These innovations are powered by Qodo’s proprietary infrastructure, including its open-source embedding model, Qodo-Embed-1-1.5B, which has demonstrated superior performance in code retrieval benchmarks compared to offerings from competitors like OpenAI and Salesforce.
Qodo is currently offering its platform under a freemium model, making it accessible to individuals for free and providing discounted rates for startups through Google Cloud’s Perks program. For larger enterprises requiring advanced controls, single sign-on (SSO), or air-gapped deployment, Qodo offers enterprise-grade solutions. The company has already established partnerships with teams at major organizations such as NVIDIA and Intuit, and thanks to a recent collaboration with Google Cloud, Qodo’s models are now available directly within Vertex AI’s Model Garden, facilitating easier integration into enterprise pipelines.
As the landscape of software development continues to evolve, the concept of context engines is poised to become a defining trend in the coming years. Qodo’s co-founder and CEO, Itamar Friedman, predicts that by 2026, every enterprise will need to develop its own “second brain” to leverage AI that truly understands and assists them. As AI systems become increasingly integrated into software development processes, tools like Qodo are demonstrating the transformative power of delivering the right context at the right moment. This shift not only enhances the efficiency of code reviews but also fundamentally changes how teams build, ship, and scale code across the enterprise.
In conclusion, the partnership between monday.com and Qodo exemplifies how innovative technology can address the challenges posed by rapid growth in software development. By harnessing the power of context engineering, monday.com has not only improved its code review process but has also fostered a culture of collaboration and continuous learning among its developers. As the demand for efficient and high-quality software delivery continues to rise, the lessons learned from this collaboration will undoubtedly serve as a blueprint for other organizations seeking to navigate the complexities of modern software development.