monday.com Rebuilt Itself Around Claude. Here Is What That Means for Your Team.

If you have added a few AI features to how your team works and quietly noticed that people tried them once and drifted back to their old habits, you are not doing anything wrong. It is the most common pattern in business AI right now, and monday.com just published a clear account of how it moved past it. Their story is worth reading whatever platform you run on, because the lesson underneath it applies at any size.

More than 250,000 companies, from small businesses to Fortune 500 organisations, use monday.com to run their work. When it launched over a decade ago, its core was a visual way to manage projects and automate workflows. In 2026 it rearchitected the product from the ground up around a human-agent collaboration model, with Claude at the core, so that AI is part of the work at every level rather than a separate feature off to the side.

monday's chief product and technology officer, Daniel Lereya, described the move as one of the most significant decisions the company has made, and framed it as reimagining what the platform should do rather than adding AI to existing workflows. The company launched the new experience in May 2026, and within about two months its customers had more than five million interactions with agents on the platform.

This is the part that matters most for the rest of us, because it names a trap that catches teams of every size.

monday first tried the obvious path: embedding AI into its existing product. That effort peaked in an internal "AI month" in May 2025, four weeks dedicated to shipping AI features across the company. Adoption looked strong at first, then hit a ceiling. The company's VP of Product for its AI Works Platform, Orly Stern Izhaki, called what they had been building "AI dust", useful features that summarised text or sorted information but never changed how the product actually created value, and never turned into habits people kept.

Her conclusion is the line worth writing on a whiteboard: adopting AI features is not the same as becoming an AI company. Once monday accepted that, it stopped decorating the old product and started rebuilding around a new idea, a place where people and agents get work done together, using the boards, permissions, and governance customers already had.

The design choice that makes this work is deceptively simple. Every monday agent gets a name, an avatar, and a place on the board, and colleagues assign it work through the same triggers and mentions they would use with a person. It has its own access permissions, just like a teammate would.

That sounds cosmetic, and it is not. It solves a pattern monday saw across its customers: plenty of companies want to use AI, but they stall at a chat window that runs parallel to where the real work happens. Putting the agent inside the workflow, where you assign it a task the way you assign one to a colleague, is what makes people actually use it without being reminded to. The jobs monday has mapped for agents run from IT ticket triage and knowledge-base upkeep to candidate screening and interview scheduling, competitive-intelligence briefings for sales and marketing, and chief-of-staff work like meeting prep and turning decisions into tracked tasks.

monday gives teams four ways to put Claude to work, and knowing them helps you pick the right one for a given job.

With monday Agents, you build a custom agent from a prompt and choose Claude as its model, and the platform gives it a name, a face, and a spot on the board. With Bring Your Own Agent, a Claude managed agent that one person builds can join the platform and become a teammate the whole team can mention and assign work to. Pre-built Agents, from the monday Agents Store, turn Claude plugins into specialised teammates, so a legal or finance team can run its own plugin as an agent inside its own workflows. And the Claude coding integration lets teams plan and assign coding tasks from the monday dashboard, where a Claude managed agent runs the work in the company's own environment and lands the result back on the item for a person to review.

One example makes it concrete. A marketing team runs a whole campaign inside a single board item. The marketer and content lead shape the brief on that item, agreeing on the goal, audience, message, and channels. A strategist agent built with monday Agents turns that into a structured brief with objectives, messaging pillars, a channel breakdown, and success metrics. A landing-page builder, running as a Claude managed agent in the company's own environment, then takes the approved brief and generates a new page variant, and the output lands back on the monday item on its own. Before it goes to approval, a brand-reviewer agent checks it against brand and legal guidelines and flags anything that needs a human. The marketing manager makes one decision: publish or refine. The work moves from brief to page without leaving the board, with a person in control at the points that matter.

monday shared five lessons, and they translate cleanly to a company a fraction of its size. The mental model was harder to change than the technology, because people naturally want to protect what already works. Small teams with clear ownership moved faster when everything changed at once. Adoption depended on trust as much as capability, so governance, permissions, and transparency decided whether agents made it past a pilot. Capability needed infrastructure to match, because agents perform better when grounded in live data and structured workflows. And the one that ties it together: build on what already works, extending the way your team already operates to a new kind of teammate rather than throwing it out.

Here is the honest bridge. You are not monday, and five million agent interactions is not a target for a team of fifteen. If you try to copy the scale, you will burn a quarter. The transferable part is not the scale, it is the diagnosis. The "AI dust" problem shows up at every size, and it is cheaper to catch early. A ten-person team that ships three AI features nobody uses has wasted three months. The same team that picks one workflow and genuinely rebuilds it around agents as teammates has something real to show. And it is not only tech giants doing this. Cooke, a family-owned seafood business founded in 1985 and now operating in 16 countries, runs project delivery and contract management on monday and Claude together, across roughly 200 projects and 130 contracts. As its director of strategy Patti Stevens put it, "Monday used to be a platform we had to update. Now we operate from it."

We are a Platinum monday.com partner and a Select partner in the Claude Partner Network Services Track, which means we work on both sides of the exact combination this story is about. We did not build monday's platform. What we do is help companies put what monday and Claude built to work, at their scale and without the trial and error.

In practice that means helping you pick the one workflow worth rebuilding first, designing agents as governed teammates inside your existing monday boards and permissions rather than as a chat off to the side, setting up the trust and governance that decide whether agents make it past a pilot, and keeping the credit and cost side predictable. The goal is the same shift monday made, scaled down to a first workflow that pays off, so your team gets the value without spending a quarter learning the hard lessons that are already written down above.

Honestly, if your use case is simple and you have someone in-house who enjoys this kind of work, start on your own. monday's own tools and templates go a long way, and there is no reason to pay for help you do not need yet. A partner earns its cost when the rebuild spans several teams, when governance and security have to hold up to a review, when the workflows carry real operational weight, or when you would rather skip the months of trial and error and start from what already works. When you reach that point, that is exactly the work we do.