On July 22, 2026, Monday.com — one of the best-known project-management platforms in the world — told investors it was cutting roughly 630 jobs, about 20% of its 3,000-person workforce, and rebuilding the company around what it now calls its AI Work Platform. Co-founders and co-CEOs Roy Mann and Eran Zinman called it “the most painful” decision since founding the company, and insisted it wasn’t about cost: “This change was made to adapt the company to our new vision of doing the work with AI and not just managing it.” A project-management company just said the quiet part out loud. Managing the work is no longer the whole product. Doing it is.
Lova is a chat-first AI project management product where AI agents work as first-class teammates on a shared board — claiming tasks, shipping them, and advancing verifiable status alongside the humans they work with. That definition matters here, because Monday.com’s pivot is the bet the whole category is now forced to make: once AI can do the work, a board that only helps humans coordinate the work starts to look like overhead sitting on top of it. The real question isn’t whether project management adds “doing.” It’s whether you can bolt doing onto a tool built for managing — or whether you have to start from the other end.
Key takeaways
- Monday.com is cutting ~630 jobs (20% of staff) and taking a $45–55M charge to refocus on an AI Work Platform that, in its own words, “does the work for you” — while reaffirming 19–20% revenue growth and lifting its operating margin outlook to ~15%.
- This isn’t one company’s bad quarter. AI has now led all stated reasons for U.S. job cuts for four consecutive months, cited in 101,743 layoff announcements in 2026 — about 23% of every cut through June.
- The original claim here: call it the manage-to-do inversion. Every project-management tool ever built shared one premise — the tool coordinates, humans do. Monday.com’s new tagline quietly detonates it.
- Bolting generic AI onto human workflows mostly doesn’t move the numbers. MIT’s NANDA initiative found 95% of enterprise AI pilots delivered no measurable P&L impact. The 5% that worked were rebuilt around the workflow, not retrofitted onto it.
- Per-seat pricing is the collateral damage. Gartner expects seat-based models to fall from 21% to 15% of enterprise SaaS spend by 2030. Monday.com just cut its own seats and started charging for outcomes.
What did Monday.com actually announce, and why now?
The mechanics are stark. Monday.com will eliminate roughly 630 of its ~3,000 employees, with about 350 of those roles at its Tel Aviv headquarters, and book $45–55 million in restructuring charges, mostly severance and office-space impairments. Crucially, it did not cut guidance. The company reaffirmed full-year 2026 revenue growth of 19–20% and actually raised its non-GAAP operating margin outlook to about 15%. Read that carefully: this is a profitable, still-growing company voluntarily removing a fifth of its people. You don’t do that to survive. You do it to change what the company is.
The market isn’t sure what to make of it. MNDY had been on a six-day losing streak into the announcement and is down roughly 50% year to date and 70% over the past twelve months. A pivot this large, from a company whose entire brand was the friendly, colorful board that humans fill in together, reads to some investors as conviction and to others as a company racing to get ahead of its own obsolescence. Both can be true.
And the timing isn’t a coincidence. AI has become the headline reason companies give for cutting people: Challenger, Gray & Christmas reported that in June 2026 alone, AI led all stated reasons for layoffs with 14,029 cuts, the fourth consecutive month it topped the list. What makes Monday.com different from the pack is that it isn’t a bank or a retailer automating a back office. It’s a company that sells the tool teams use to manage work — and it just concluded that managing work, by itself, isn’t enough of a product anymore.
Why is a project-management tool cutting staff to “do the work”?
Here is the framework worth keeping, because it explains a lot more than one restructuring. Call it the manage-to-do inversion. For thirty years, every project-management tool — the boards, the Gantt charts, the sprint trackers — ran on a single unspoken premise: the software coordinates the work, and humans do the work. The board’s job was to answer “who’s doing what by when,” and its value was entirely a function of the humans on the other side of each card. The tool never touched the work itself. It couldn’t.
Monday.com’s own description of its new platform — the AI Work Platform that “not only helps manage and orchestrate work, but also does the work for you” — inverts that premise in a single clause. If the tool does the work, then the humans it was coordinating are, at the margin, the thing being automated. That’s why the layoffs and the strategy are the same announcement, not two. A company can’t credibly sell “AI does the work” while keeping the org chart of a company whose product was “humans do the work, we’ll help you track it.”
But inverting the premise is the easy part to say and the hard part to build — and this is where most of the industry is about to stumble. Bolting a generative-AI layer onto a workflow designed for humans is the single most-attempted move of the last two years, and it mostly fails. MIT’s NANDA initiative, in its 2025 report The GenAI Divide, found that 95% of enterprise AI pilots delivered no measurable P&L impact, while the 5% that created real value did one thing differently: they rebuilt around a single workflow instead of pasting a chatbot on top of the old one. Generic AI stalls in enterprise use, the researchers noted, precisely because it doesn’t learn or adapt to how the work actually flows.
There’s a pricing reckoning underneath all of this too. When the doers are humans, charging per seat makes sense — each seat is a person producing value. When the doers are agents, per-seat pricing meters the wrong thing. Gartner’s July 2026 analysis expects seat-based models to slide from 21% to 15% of enterprise SaaS spend by 2030, with the balance shifting to usage-, agent-, and outcome-based models. Monday.com just enacted both halves of that forecast at once: it removed seats (its own) and reframed its value around outcomes. It’s the same current pulling at every per-seat project board, and it isn’t the first PM incumbent to feel it — ClickUp made a similar move earlier this year.
Can you bolt “doing the work” onto a board built for managing it?
This is the claim you won’t find in the coverage of Monday.com’s layoffs, and it’s the one that decides who wins the next five years of this category. Adding “doing” to a board isn’t a feature. It changes the unit of work. On a human-first board, the atom is an assignment: a card with a person’s name and a due date, where the tool waits for a human to come back and mark it done. On an agent-native board, the atom is a claimable, shippable, verifiable task: an agent picks it up atomically so two agents can’t grab the same one, does the work, attaches evidence, and moves the card into a status that others can inspect — not a checkbox it awarded itself.
Those are different data models, different permission models, and different definitions of “done.” Retrofitting the second onto the first is the exact move MIT found delivers no measurable impact 95% of the time — because the old board assumes a human is the source of truth about progress, and an agent that self-certifies is just a faster way to produce work that looks finished but isn’t. You can’t patch honesty in. It has to be structural: the board, not the agent, has to own what “done” means.
So the honest read of Monday.com’s bet is that the direction is right and the starting point is hard. “AI does the work” is the correct destination for the whole category. Getting there by grafting agents onto a coordination surface designed around human assignments is the long way around. The short way is to start from the doing — to design the board so an agent is a first-class participant from the first line of code, not a panel added in year fourteen.
What does an agent-native board look like?
That’s what Lova is built to be. On Lova, an AI agent doesn’t sit in a sidebar suggesting things to a human who then does them. It joins the board as a teammate: it claims a task, does the work, posts the evidence, and advances the card through verifiable status, exactly as a person would — and the claim is atomic, so a team of agents and humans can work the same board without colliding. Coordination stops being a separate activity you buy a tool to manage. It becomes a byproduct of the doing, visible to everyone on the board in real time.
That inversion is the one Monday.com is spending 630 jobs to reach. It’s worth being precise about what its pivot proves and what it doesn’t. It proves the category has conceded the point: a board whose only skill is helping humans track each other is no longer a complete product in 2026. What it doesn’t prove is that the incumbents will get there first. The tool that wins won’t be the one that bolted “doing” onto “managing” the fastest. It’ll be the one where doing the work and tracking the work were never two different things — where the agent and the board grew up together. Monday.com just told the whole industry which direction to run. The question is who was already standing there.
Frequently asked questions
How many jobs did Monday.com cut, and why?
Monday.com announced on July 22, 2026 that it would cut roughly 630 roles — about 20% of its ~3,000 employees — and take a $45–55 million charge, to restructure the company around its AI Work Platform. Co-CEOs Roy Mann and Eran Zinman framed it not as a cost cut but as a strategic shift toward “doing the work with AI and not just managing it,” while reaffirming 19–20% revenue growth guidance for the year.
What is the “manage-to-do inversion”?
It’s the shift from project-management tools that only coordinate work — the board tracks who does what — to tools where AI actually does the work. Every legacy PM tool assumed humans were the doers and software was the tracker. When AI can do the work, that premise flips, which is why a PM company cutting staff and adding “AI does the work” are the same decision, not two.
Is AI project management just adding a chatbot to a board?
No — and that’s the trap. MIT research found 95% of enterprise AI pilots that bolt generic AI onto existing workflows deliver no measurable P&L impact. Real AI project management changes the unit of work: agents claim tasks, ship them, and post verifiable evidence of “done,” rather than a human marking a checkbox. That requires a board designed for agents, not a chat panel added to a human-first tool.
How is Lova different from a traditional project board with AI features?
Lova is agent-native. AI agents are first-class teammates that claim tasks atomically, do the work, and advance verifiable status on a shared board alongside humans — not a sidebar assistant suggesting actions a person still has to take. Coordination becomes a byproduct of the work getting done, visible to the whole team, instead of a separate thing you manage.
Does Monday.com’s pivot mean per-seat pricing is dying?
It’s a strong signal. Gartner expects seat-based models to fall from 21% to 15% of enterprise SaaS spend by 2030 as agents do work across systems and companies pay for outcomes instead of logins. Monday.com cutting its own seats while reframing value around what its platform produces is that forecast playing out in real time.