monday.com запустила AI-агентов в продакшене на Amazon Bedrock: +50% PR на инженера

monday.com has rolled out its agentic AI Teammates into production on the Amazon Bedrock platform and disclosed internal metrics on their performance: nine out of ten of the company's developers use AI coding tools every month, and pull request output per engineer has grown by more than one and a half times.

What numbers did monday.com reveal?

monday.com cites metrics from its own production analytics, and all of them point to mass adoption of AI in development, not a pilot. Six months ago, about half of engineers regularly used AI coding tools — now it's 9 out of 10.

• Nine out of ten developers ("Builders") use AI coding tools monthly — up from roughly half six months ago
• Pull request output per engineer has grown by more than 50%
• The agents run on Amazon Bedrock — AWS's managed service for running and orchestrating large language models
• All figures come from monday.com's internal production data, not from a survey or marketing report

According to the AWS Machine Learning Blog, monday.com is one of the few engineering organizations running agentic AI in production at this scale, rather than as a demo.

How the agents were built into a 10-year-old codebase

The main engineering challenge wasn't the agents themselves, but fitting them into a codebase roughly a decade old. Such projects accumulate legacy patterns, non-standard conventions, and dependencies that trip up general-purpose AI tools.

monday.com describes this as "retrofits" — modifications that made the existing architecture suitable for agent work: codebase context, guardrails, and integrations, without which the model produces code that is formally correct but useless in this particular project. The company names these retrofits, not the choice of a specific model, as the reason the setup worked at scale.

What is confidence-scored merge?

Confidence-scored merge is the technique monday.com uses to narrow the gap toward full agent autonomy. The system assigns a confidence score to every proposed change and uses that score to decide whether to merge the code automatically or send it to a human for review.

This approach is a trade-off between speed and control: routine, predictable edits with high confidence go through without human involvement, while anything questionable stays under manual control. It's the very last line separating an assistant that writes code under supervision from an agent that is trusted to merge on its own.

Every figure in this piece comes from monday's own internal production data — as stated on the AWS

Machine Learning Blog.

What this means

monday.com's case is a rare example of agentic AI in development being measured not by promises but by production metrics: share of users, PR output, share of automatic merges. For teams with large legacy projects, the takeaway is simple — the bottleneck isn't the model, but what's built around it: context, guardrails, and rules for automatic merging.

Frequently asked questions

What is AI Teammates?

AI Teammates is monday.com's agentic AI, built on Amazon Bedrock. Unlike a suggestion-based assistant, these agents carry out development tasks and take changes all the way to a pull request, merging some of the edits automatically based on a confidence score.

What platform do monday.com's AI agents run on?

monday.com's agents run on Amazon Bedrock — AWS's managed service for running and orchestrating large language models. It's the platform on which the company runs production agents at the scale of its entire engineering organization.