ClickUp Released 290 Agent Operators on May 21. Asana Just Bought the Rails.

On May 21, 2026 at 7:34 PM, Zeb Evans posted a memo to his 229,600 X followers announcing ClickUp had cut roughly 290 of its 1,300 employees, skipped the all-hands, and rebuilt the survivors' comp around a 3:1 agent-to-human ratio with cash bands reaching $1 million. The cohort that walked out is the most agent-fluent operator pool currently reachable on the open market. Asana, Monday, and Notion have about 45 days to absorb them before the survivor screenshots hit X and reset comp expectations across every SaaS productivity company in the bracket.

What Evans actually did

The number is 22%. The framing is the story. Evans did not call this a rightsizing, a refocus, or a path to profitability. He called it the front edge of a transition every SaaS company will make, and he tied retained roles to a specific operating model: roughly 3,000 internal AI agents running across engineering, marketing, support, and exec ops, with humans reorganized into three categories.

Builders are engineers and PMs who direct agents instead of writing code. System Managers automate their own workflows and supervise the agents downstream. Front-liners stay in direct customer interactions. Everyone else got severance.

The comp piece is what matters for sourcing. Evans published $1M cash bands (not equity) for survivors who produce outsized output with AI. He did this on his personal account, with his name attached, on a Wednesday night, which means he wants the screenshots to spread. He is using the memo to recruit the next wave at his price, and to make sure every competitor's staff engineer knows what the ceiling looks like.

The Codegen tell

Late 2025, ClickUp acquired Codegen, the AI coding platform. That acquisition is why the ClickUp engineering org was agent-fluent six months before anyone else's. Fortune reported the 3,000-agent figure on May 18. The layoffs hit on the 21st. The internal deployment had been running since January 2026, which means the cut cohort spent four to five months training, supervising, and handing off production workflows to agents that ultimately displaced them.

That is the part recruiters keep missing. The survivors got the credit. The released cohort did the agent-onboarding labor.

Who actually got cut

Most layoff-recruiting playbooks default to "source the engineers." For ClickUp 290, that is the wrong filter.

The rarer profile is the growth ops manager, the IT ops lead, the support automation owner. Business Insider named Andy Cabasso, a ClickUp growth operations manager who oversees 37 AI agents. He is the archetype. PMs and ops people who built the eval criteria, the prompt scaffolding, the exception-handling routines, the human-in-the-loop checkpoints. They have shipped what Asana's AI Teammates and Monday's Digital Workforce sell to customers, except they have done it at production scale on a live SaaS company's revenue motion.

37

There is no public credential for this skill yet. No certification, no widely indexed job title, no "Agent Operator II" on LinkedIn. Which is exactly the sourcing problem. You cannot keyword-match your way to these people because the keywords do not exist. You have to describe the work in plain English: "ex-ClickUp ops or PM who managed 20+ internal agents in production between January and May 2026." That is the kind of query Refolk was built for, because it parses the description against GitHub commits, LinkedIn history, conference talks, and the open web instead of fighting a Boolean string against a saturated index.

Why Asana is the obvious buyer and the obvious miss

Seven days after ClickUp cut 290 people, Asana announced it had acquired Stack AI, the no-code agent builder, for a reported $75 million. CEO Dan Rogers has been using the phrase "operating system for human-agent teams" since AI Teammates went GA in April 2026 at $15 per user per month.

Read those two press releases side by side. Asana spent $75M on the rails and did not budget for the conductors. The Stack AI founders (Tony Rosinol and Bernard Aceituno) are joining to build the infrastructure. Nobody at Asana has run 37 production agents on a live customer base for five months. ClickUp's growth ops team has. The arbitrage is sitting in plain sight, and Asana's stock at $6.88 (down 53% from January) means cash offers within reasonable bands will close.

Asana bought the rails. ClickUp released the conductors. The trade closes itself if you move in 45 days.

Notion runs 2,800 agents and doesn't talk about it

Notion disclosed in February that early testers had built more than 21,000 Custom Agents, and that Notion itself runs 2,800 internal agents around the clock. That number is almost identical to ClickUp's 3,000. The tooling fluency overlaps. The scale overlaps. Notion has not published a headcount ratio because Notion does not need to recruit through memo theater, but structurally it is the closest absorber of the ClickUp cohort.

The Notion Custom Agents launch named operators like Ben Levick at Ramp and James Lawley at Remote as production agent users. That is the comparison class. A ClickUp ops manager running 37 agents internally walks into Notion as a senior IC who already speaks the product. The hire should be quiet and immediate.

Monday.com's Digital Workforce needs operators, not marketing

Monday markets its Digital Workforce as autonomous agents that handle routine work 24/7 and learn from user behavior. The product exists. The headcount that knows how to deploy it on a live ARR base does not. ClickUp's cut included CS and support automation owners who spent five months designing the exception paths Monday's customers will hit in week one of any real rollout.

Monday's recruiting team will default to engineer titles. That is the wrong filter again. The trade here is on system-manager profiles: people who automated their own workflow first, then their team's, then their department's. They are visible by what they shipped, not what their title says. Sourcing AI-fluent engineers is the easier search; sourcing AI-fluent operators is where the alpha is, and it is precisely where most ATS pipelines fail.

The 45-day window is a comp clock, not a recruiting clock

The conventional read on a layoff-recruiting window is "you have N days before they take another job." For ClickUp 290, the timer is different. The window is the time before ClickUp survivors start posting $1M comp screenshots on X.

Evans optimized the memo for this. Public, on his personal account, his name attached, $1M cash (not equity), framed as the new ceiling for "outsized impact using AI." Once two or three survivors post their offer letters, the floor for AI-fluent staff engineer and senior ops comp at Asana, Monday, and Notion resets industry-wide. Every existing offer in market gets renegotiated. Every counter-offer at the survivor company gets richer. The displaced cohort, who are currently negotiating at pre-reset bands, will get pulled up with the market.

Sourcing the cohort before that happens is the only way to hire them at current comp. After the reset, you are paying for the same skill at the new ceiling, against survivors who are now public about their number.

$1M

Wait, that block should not have a stray closer. Treat the stat block as ending after the note.

Why LinkedIn is not the tool for this cohort

Refolk's index shows the publicly searchable ex-ClickUp pool was still extremely thin across product, engineering, and design titles in early June. That is not a bug in LinkedIn. It is the timeline. Severance paperwork, NDAs, and the standard 30-to-60-day window before people switch their headline mean the cohort is invisible to keyword search exactly when it is most reachable on email and DM.

This is the structural reason agent-to-human ratio hiring breaks LinkedIn Recruiter. The skill has no keyword. The cohort has no headline change yet. The relevant signal is in commit history (Codegen integration work), Loom-style demo posts on X, internal-tooling talks at small conferences, and the comment thread under Evans's memo itself. Pulling that into a shortlist is what Refolk does in one query instead of forty Boolean variations.

The Codegen filter

The single highest-signal sourcing filter inside the ClickUp 290 is "touched Codegen integration." Codegen was acquired late 2025 and embedded into ClickUp's engineering workflow before the January 2026 internal rollout. Engineers who worked that seam are the only people on the open market who have shipped agent-directed code at SaaS production scale on a recently acquired platform. There are not many of them. They will not surface on a title search. They will surface on a description-based query that knows what Codegen is.

What to do this week

If you run talent at Asana, Monday, Notion, Coda, Airtable, Smartsheet, or any of the second-tier productivity SaaS players: pull a list now, before mid-July.

The list is not "ex-ClickUp." The list is three slices. First, growth and ops managers who publicly referenced agent counts in the 10 to 40 range on X or in Notion's customer stories. Second, PMs who shipped between January and May 2026 on internal AI tooling, with a Codegen filter on the eng side. Third, support and CS automation owners whose LinkedIn says "Customer Success" but whose actual job was deploying autonomous workflows on the support queue.

Skip the recruiter screen for week one. Send a founder or VP note that names the work. "Saw you ran the agent rollout on the support side. We bought the rails and need a conductor. Cash band is X, here is what closes this week." That is the offer Zeb Evans is daring you to make. He published the comp ceiling on X so the market would chase it. The trade only works if you move before the market catches up.

FAQ

How many people did ClickUp actually cut on May 21, 2026?

Roughly 290 employees, or about 22% of the company's 1,300 headcount. The cut was announced by CEO Zeb Evans on his personal X account at 7:34 PM Pacific on May 21, 2026, and first reported by TechCrunch's Marina Temkin on May 25. The memo tied retained roles to a 3:1 agent-to-human ratio across 3,000 internal AI agents and introduced $1 million cash compensation bands for survivors producing outsized output.

Why are growth ops and system managers more valuable than engineers in this cohort?

Because Asana, Monday, and Notion already have engineers who can build agents. What they do not have is people who have run agents at production scale on a live SaaS ARR base for five months. Andy Cabasso, named in Business Insider, oversaw 37 agents as a ClickUp growth ops manager. That archetype (eval criteria design, prompt scaffolding, exception handling) is the rarer and harder-to-credential skill, and it does not surface on a title-based LinkedIn search.

Why is the window 45 days specifically?

It is a comp-reset clock, not a recruiting clock. Evans published the $1M cash band on a public X post optimized for screenshot virality. Once ClickUp survivors start posting their offer letters publicly, the comp floor for AI-fluent staff engineer and senior ops roles at every comparable SaaS company resets upward. Hiring the displaced cohort before that happens is the only way to close offers at current bands.

What is the highest-signal filter inside the ClickUp 290?

For engineering, it is exposure to the Codegen integration. ClickUp acquired Codegen in late 2025 and embedded it into the internal agent rollout that began January 2026. Engineers who worked that seam shipped agent-directed code at production SaaS scale earlier than anyone else on the open market. For non-engineering, it is operators who publicly referenced agent counts in the 10 to 40 range or appear in customer stories from Notion, Stack AI, or comparable tooling vendors.