How context engineering can save your company from AI vibe code overload: lessons from Qodo and Monday.com
As a cloud software for project tracking monday.comAs the engineering organization surpassed 500 developers, the team began to feel the pressure of its own success. Product lines proliferated, microservices proliferated, and code flowed faster than human reviewers could keep up. The company needed a way to review thousands of pull requests every month without drowning developers in boredom – or sacrificing quality.
That’s when Guy Regev, VP of R&D and head of the Growth and Monday Dev teams, started experimenting with a new AI tool from Qodoan Israeli startup focused on developer agents. What started as a lightweight test quickly became a critical part of monday.com’s software delivery infrastructure new case revealed by both Qodo and monday.com.
“Qodo doesn’t feel like just another tool – it’s like adding a new developer to the team who actually learns how we work,” Regev told VentureBeat in a recent video call, adding that it “prevented more than 800 issues per month from entering production – some of which could have caused serious security issues.”
Unlike code generation tools like GitHub Copilot or Cursor, Qodo doesn’t attempt to write new code. Instead, it specializes in assessing them – using what it calls context technique to understand not only what has changed in a pull request, but why, how it fits the business logic, and whether it follows internal best practices.
“You can call Claude Code or Cursor and in five minutes you’ll get 1,000 lines of code,” says Itamar Friedman, co-founder and CEO of Qodo, on the same video call as with Regev. “You have 40 minutes and you can’t watch that. So you need Qodo to actually watch it.”
For monday.com, this opportunity was not only useful, but transformative.
Code review, at scale
At any given time, monday.com developers send updates to hundreds of repositories and services. The engineering organization works in tightly coordinated teams, each tailored to specific parts of the product: marketing, CRM, development tools, internal platforms and more.
That’s where Qodo came into the picture. The company’s platform uses AI not only to check for obvious bugs or style violations, but to evaluate whether a pull request follows team-specific conventions, architectural guidelines, and historical patterns.
It does this by learning from your own codebase: training on past PRs, comments, merges, and even Slack threads to understand how your team works.
“The comments Qodo provides are not general: they reflect our values, our libraries, and even our standards for things like feature flags and privacy,” Regev said. “It’s context-aware in a way that traditional tools aren’t.”
What “Context Engineering” actually means
Qodo mentions his secret sauce context technique – a systems-level approach to managing everything the model sees when making a decision.
This obviously includes the PR code diff, but also previous discussions, documentation, relevant files from the repository, even test results and configuration data.
The idea is that language models don’t really ‘think’: they predict the next token based on the input they receive. The quality of their output therefore depends almost entirely on the quality and structure of their input.
As Dana Fine, Qodo’s community manager, puts it in one blog post: “You don’t just write prompts; you design structured input under a fixed token limit. Every token is a design decision.”
This is not just theory. In the case of monday.com, this meant that Qodo could catch not only the obvious bugs, but also the subtle bugs that typically slip past human reviewers: hardcoded variables, missing fallbacks, or violations of architectural conventions between teams.
One example stood out. In a recent PR, Qodo highlighted a line that inadvertently exposed a staging environment variable – something no human reviewer noticed. If it had been merged it could have potentially caused problems in production.
“The hours we would spend resolving this vulnerability and the legal issues it would pose would be far more than the hours we would save after a pull request,” Regev said.
Integration into the pipeline
Today, Qodo is deeply integrated into monday.com’s development workflow, analyzing pull requests and making context-aware recommendations based on previous team code reviews.
“It doesn’t feel like just another tool… It feels like a new teammate who has joined the system – someone who is learning how we work,” Regev noted.
Developers receive suggestions during the review process and retain control over the final decisions – a human-in-the-loop model that was critical to adoption.
Because Qodo integrated directly into GitHub via pull-request actions and comments, Monday.com’s infrastructure team didn’t have a steep learning curve.
“It’s just a GitHub action,” Regev said. “It creates a PR with the tests. It’s not a separate tool that we had to learn.”
“The goal is to actually help the developer learn the code, take ownership, give each other feedback, learn from it and set the standards,” Friedman said.
The results: time saved, bugs prevented
Since Qodo rolled out more broadly, monday.com has seen measurable improvements across multiple teams.
Internal analyzes show that developers save an average of about an hour per pull request. Multiply that by thousands of PRs per month, and the savings quickly reaches thousands of developer hours per year.
These aren’t just cosmetic issues; many of these have to do with business logic, security, or runtime stability. And because Qodo’s suggestions reflect monday.com’s actual conventions, developers are more likely to act on them.
The system’s accuracy is rooted in its data-first design. Qodo trains on each company’s private codebase and historical data, adapting to different team styles and practices. It does not rely on one-size-fits-all rules or external data sets. Everything is custom.
From internal tool to product vision
The Regev team was so impressed with Qodo’s impact that they started planning deeper integrations between Qodo and Monday Dev, the developer-focused product line monday.com is building.
The vision is to create a workflow where the business context – tasks, tickets, customer feedback – flows directly to the code review layer. This way, reviewers can assess not only whether the code ‘works’, but also whether it solves the right problem.
“We used to have linters, hazard rules, static analyses… rules based… you have to configure all the rules,” Regev said. “But it doesn’t know what you don’t know… Qodo… feels like it’s learning from our engineers.”
This is closely aligned with Qodo’s own roadmap. The company doesn’t just review code. It builds a full platform of developer agents, including Qodo Gen for context-aware code generation, Qodo Merge for automated PR analysis, and Qodo Cover, a regression testing agent that uses runtime validation to ensure test coverage.
All of this is made possible by Qodo’s proprietary infrastructure, including its new open-source embedding model, Qodo-Embed-1-1.5B, which outperformed OpenAI and Salesforce offerings on code retrieval benchmarks.
What’s next?
Qodo now offers its platform under a freemium model: free for individuals, discounted for startups through Google Cloud’s Perks program, and at an enterprise level for companies that need SSO, air-gapped deployment, or advanced controls.
The company already works with teams at NVIDIA, Intuit and other Fortune 500 companies. And thanks to a recent partnership with Google Cloud, Qodo’s models are available directly in Vertex AI’s Model Garden, making it easier to integrate into business pipelines.
“Context engines will be the big story of 2026,” Friedman said. “Every company will have to build its own second brain if they want AI that actually understands and helps them.”
As AI systems become increasingly embedded in software development, tools like Qodo show how the right context – delivered at the right time – can transform the way teams build, ship, and scale code across the enterprise.
