ClickUp lays off staff en masse to bet on AI agents

ClickUp, a textbook case of the shift from copilots to agents

The signal is brutal, and it goes far beyond the case of a single company. According to TechCrunch, in an article titled “What ClickUp’s mass layoff tells us about the future of work”, U.S. software publisher ClickUp reportedly eliminated hundreds of positions while simultaneously deploying thousands of AI agents internally. The information, even if it was not presented by the company as a conventional product announcement, acts as a revealer of a phase change in the software industry: generative AI is no longer used only to assist employees, it is beginning to directly reorganize team structures.

ClickUp is not an insignificant startup. Founded in 2017 and positioned in the highly competitive work management tools market, the company has established itself as one of the most visible names in the “collaborative productivity” segment, alongside Asana, Monday.com, Notion, Atlassian, and Microsoft’s suites. Its ambition was clear from the outset: to offer a unified platform for tasks, documents, goals, discussions, and automations, with a promise of efficiency aimed at teams of all sizes. In this type of software, value is not limited to the interface. It also depends on internal operations: customer support, onboarding, marketing, sales, documentation, quality, customer success, operational finance. Yet these are precisely the functions that AI agents are beginning to affect.

The ClickUp case is of particular interest to French-speaking audiences because it gives concrete form to a fear often debated in the abstract: what happens when “copilots” become “agents” capable of executing complete tasks, in sequence, with reduced supervision? Since late 2022, generative AI was first sold as a tool for augmenting human work. The market’s vocabulary was carefully chosen: assistant, copilot, writing aid, productivity booster. Two years later, the lexicon is changing. People now talk about agents, process automation, digital workforce, systems capable of handling entire operations. The difference is not merely marketing. It touches the very definition of a job.

For SaaS vendors, this transition is almost structural. These companies sell software meant to improve their customers’ productivity, but they are also themselves machines built on standardized processes. They produce immense volumes of support tickets, marketing content, sales follow-ups, meeting summaries, documentation, internal workflows, customer responses, lead qualification, and reporting. Anything repetitive, measurable, textual, or semi-structured is now a natural target for AI agents. In this context, the story reported by TechCrunch does not appear as an anomaly, but as an advanced prototype of what part of the software sector could become.

ClickUp must also be placed within a broader moment in the industry. Since 2023, major technology players have multiplied announcements around agents. OpenAI is pushing ChatGPT toward task execution and tool orchestration. Microsoft is integrating Copilot into its professional suite and developing business-agent scenarios. Salesforce launched Agentforce with the explicit idea of creating a digital workforce for customer relations. Google is equipping Workspace and Cloud with agentic functions. ServiceNow, HubSpot, Zendesk, SAP, and Oracle are all moving in the same direction. What is new in the ClickUp case is that we are no longer talking about a commercial demo or a keynote, but about a concrete HR effect: job cuts associated with a ramp-up in internal automation.

For Europe and France, where the AI debate is often structured around regulatory, ethical, and sovereignty issues, this case adds a more social and more immediate dimension. French companies have so far largely adopted generative AI in the form of experiments, pilots, and assistance tools. The shift to agents capable of replacing part of administrative, commercial, or support work raises a different question: no longer “what can AI do?” but “which jobs will be reshaped first?” The ClickUp case shows that the answer could come faster than expected in software companies, and then spread to other service sectors.

What TechCrunch reports: job cuts and an army of AI agents

According to TechCrunch AI, ClickUp reportedly carried out layoffs affecting hundreds of employees, while simultaneously deploying thousands of AI agents to operate internally. That is the core of the information, and it is strong enough to make this sequence much more than a simple cost-cutting plan. The American outlet sees it as an indicator of the evolution of office work, particularly in digital companies where information flows are already largely digitized and therefore easier to automate.

The essential point is not only the scale of the job cuts, significant though it is. It is the claimed or accepted correlation between human restructuring and operational substitution by agents. In the waves of layoffs seen since 2022 in tech, companies often cited post-pandemic correction, budget discipline, rising rates, or slower-than-expected growth. Here, AI appears as an active variable in the reorganization. That changes the nature of the message sent to the market, employees, and investors.

The term “AI agents” deserves clarification. This is not just about chatbots capable of answering questions. In the industry’s current vocabulary, an agent is generally a system that combines a language model, contextual memory, access to tools or databases, and the ability to chain several steps together to achieve a goal. An agent can classify requests, query a CRM, generate a response, adapt it to the company’s tone, trigger an action in another piece of software, then report back with a summary. Where a copilot assists a human with a task, an agent takes charge of part of the workflow.

In a company like ClickUp, the potential uses are numerous. Customer support can be automated for recurring requests, with escalation to a human for complex cases. Marketing can generate content variants, analyze campaign performance, prepare briefs, and segment audiences. Sales can rely on agents for qualification, follow-up, and meeting preparation. Product teams can automate the synthesis of user feedback and the production of documentation. Internal functions, from finance to human resources, can also be affected through workflows for document analysis, control, or coordination.

What stands out in TechCrunch’s account is the scale mentioned: thousands of agents. Even if that expression may cover specialized agents, automated workflows, or software instances assigned to different tasks, the number suggests advanced industrialization. We are no longer talking about a few experimental assistants used by a handful of executives. We are talking about a distributed work infrastructure, potentially omnipresent throughout the organization. This points to a logic already visible among some enterprise AI providers: each team, each function, or even each employee could have one or more dedicated agents.

ClickUp’s choice is all the more symbolic because the company itself sells a promise of organizational efficiency. If a productivity software vendor reorganizes its own operations around agents, it sends an implicit message to its customers: what we sell to you, we apply to ourselves. In the SaaS economy, this form of “dogfooding” has always had strategic value. But with AI, it can become socially explosive, because it involves demonstrating real productivity gains, sometimes at the cost of reducing headcount.

Still, journalistic caution is needed in interpreting this. A mass layoff is never reducible to a single cause. In tech, it can result from a stack of factors: commercial slowdown, investor pressure, the need to improve margins, overstaffing inherited from a period of hypergrowth, reallocation toward new products. What is original about the ClickUp case is that AI is no longer a secondary context, but a central element of the narrative. Even if it does not explain everything, it becomes a credible operational justification for redefining the size and composition of teams.

For labor market observers, this point is crucial. Since the arrival of ChatGPT in late 2022, many studies have measured individual productivity gains: faster writing, better synthesis, coding assistance, document preparation, improved customer relations. But the shift from individual gain to the managerial decision to cut jobs had remained relatively diffuse, often masked by other economic motives. The ClickUp case gives this translation a more concrete face. It shows how a company can consider that automation is no longer merely support for its teams, but a new organizational foundation.

From assistance to substitution: why agents change the nature of automation

To understand the significance of the ClickUp case, we need to go back to the recent evolution of generative AI tools. The first wave, between late 2022 and 2023, was dominated by conversational interfaces and writing functions. The user asked a question and got a text, a summary, an outline, a code suggestion. The model remained in a relatively simple relationship with the human: it proposed, the human validated. This logic fed the idea of AI as a “copilot,” that is, a system that accompanies without substituting.

The second wave, initiated in 2024 and accelerated in 2025, is based on deeper integration of models into information systems. Agents no longer merely generate a response. They can access business applications, trigger actions, follow rules, retain context, collaborate with other agents, and operate asynchronously. The economic leap is considerable. A tool that helps an employee respond faster is useful. A system that handles on its own 60%, 70%, or 80% of a standardized flow changes the cost structure.

This difference explains why the most exposed functions are not necessarily the most “intellectual” in the classical sense, but those where value is produced through sequences of repeatable operations. Level 1 support, sales qualification, certain customer success tasks, document management, reporting, ticket analysis, administrative coordination, or large-scale content production are particularly affected. In these areas, the agent does not need to be “intelligent” in the human sense. It needs to be sufficiently reliable, fast, inexpensive, and well connected to data.

SaaS vendors are ideal ground for this shift for at least four reasons. First, their processes are already heavily instrumented: CRM, helpdesk, analytics, marketing automation tools, knowledge bases, project management, documentation, internal messaging. Second, their data is mostly digital and often structured or semi-structured. Third, their culture is generally favorable to rapid software experimentation. Finally, their investors expect measurable gains in margin and productivity. In this environment, agents become a logical extension of the operating model.

This dynamic does not concern startups alone. Salesforce, for example, has made Agentforce one of the main pillars of its recent strategy, with the idea that a company will be able to deploy agents to manage customer relations, sales, or service. Microsoft is pushing a similar vision with Copilot Studio and agents integrated into Microsoft 365, Dynamics, and Power Platform. ServiceNow is betting on the automation of complex workflows. Zendesk, Intercom, and Freshworks are also positioning themselves around increasingly autonomous forms of customer support. The difference is that these groups mainly communicate about the value created for their customers. The ClickUp case shows what that can look like internally.

There is also a decisive financial dimension. The marginal cost of an AI agent remains variable depending on the models used, API calls, integrations, and the human supervision required. But in many cases, it becomes competitive against the total cost of an employee for standardized tasks, especially in companies operating at international scale. An agent can run continuously, absorb spikes in demand, handle multiple languages, and produce fine-grained traceability. If its quality is judged sufficient, the economic calculation becomes tempting. This calculation does not mean the agent replaces an employee identically. It means the company can redesign the role to retain only the high-value or high-responsibility tasks.

The semantic shift observed in tech is revealing in itself. For a long time, executives spoke of “augmenting teams.” Now, they more readily refer to “AI-native organizations,” “lean teams,” or a “blended workforce” combining humans and agents. Behind these formulas lies a simple hypothesis: some companies will be able to generate as much, or even more, revenue with fewer direct employees. This promise appeals to investors, but it raises a major social question: which jobs remain central when mass execution is entrusted to software systems?

The ClickUp case suggests a partial answer. The roles that remain strategic are those that define objectives, control quality, manage exceptions, arbitrate priorities, design processes, negotiate with key customers, bear legal responsibility, or build the product. By contrast, positions centered on the repetitive execution of informational processes appear more vulnerable. That does not mean their total disappearance, but a potential compression in the number of people needed to deliver the same service.

For the French-speaking market, this distinction is important. In France, many companies first adopted generative AI for writing or exploratory uses. The shift to agents truly connected to internal tools remains slower, notably for reasons of security, compliance, and governance. But this caution may only temporarily delay the movement. Once guardrails are in place, European companies could follow the same trajectory as American players, with a lag of a few quarters. The ClickUp precedent then serves as a real-world test for anticipating the trade-offs to come.

ClickUp versus other SaaS players: what this sequence reveals about the market

To measure the significance of the case, it must be compared with competing strategies. Over the past eighteen months, almost all collaborative work software vendors have added an AI layer to their offering. Notion introduced writing, search, and assistance functions into its workspaces. Atlassian integrated AI into Jira and Confluence. Asana developed automation features and summary generation. Monday.com is pushing AI scenarios for workflows. HubSpot added many generative tools for marketing and sales. In most cases, official communication emphasizes productivity gains for customers, not the transformation of internal headcount.

The difference with ClickUp, as TechCrunch tells it, is that the vendor seems to embrace massive use of agents as a reorganization lever. That places the company in a particular category: no longer just sellers of AI tools, but companies that themselves become demonstrators of the AI-optimized-headcount enterprise. Yet this position is ambivalent. On one hand, it may reassure some investors, who see in it management capable of quickly extracting productivity gains. On the other, it may worry employees, customers, and candidates, especially if service quality deteriorates or internal culture is weakened.

The comparison with Salesforce is illuminating. The group led by Marc Benioff has multiplied public statements around the idea of a “digital labor revolution.” Benioff has publicly explained that AI would make it possible to change the way companies manage their teams, particularly in customer service and commercial functions. Salesforce has even put forward ambitious figures on the potential of its agents to handle a large number of interactions. But at this stage, communication remains largely customer-facing. The ClickUp case offers a starker glimpse of what this revolution can mean for employment at the vendor itself.

Microsoft, for its part, is taking a more institutional posture. The group presents Copilot as a cross-functional productivity tool, enriched with connectors, agents, and automation via Power Platform. The demonstrations show employees assisted in their daily tasks, with an emphasis on security, governance, and integration into the existing professional environment. The message is less direct about workforce reduction, even if the long-term economic effect may be similar. Microsoft’s strategy is to make the agent acceptable in the enterprise by first placing it under the banner of assistance. ClickUp, by contrast, seems to illustrate a more advanced phase, where assistance leads to restructuring.

Specialized customer service players such as Intercom, Zendesk, or Freshworks must also be watched. These companies were among the first to claim high ticket automation rates thanks to AI. Support is indeed one of the areas where return on investment is fastest: high volume, recurring requests, high human costs, abundant historical data. If ClickUp did in fact deploy thousands of agents, it is likely that functions close to these were among the first affected. That matches market logic: start with the flows where automation is most measurable.

Another point of comparison concerns the relationship between growth and headcount. During the 2010s, many SaaS companies were valued on their ability to grow fast, even if that meant hiring massively in sales, marketing, and support to sustain expansion. Since the macroeconomic reversal of 2022, investors have placed greater emphasis on capital efficiency, gross margin, free cash flow, and operational discipline. In this new framework, AI plays the role of an accelerator of “rationalization.” A company capable of maintaining growth while reducing labor costs can quickly improve its financial indicators. The ClickUp case fits fully within this post-ZIRP logic, that is, after the period of abundant money and boosted valuations.

For European companies, the temptation to imitate this model will be strong, but with additional constraints. Labor law, consultation obligations, data protection, and social culture make mass layoffs more complex than in the United States. In France, for example, a reorganization explicitly justified by AI automation would immediately raise legal, union, and political questions. That does not block the movement, but it can slow it down, make it more gradual, or shift it toward other mechanisms: hiring freezes, non-replacement of departures, outsourcing, role redefinition, selective upskilling.

The market could therefore segment. American players, faster in adoption and more flexible socially, would serve as the laboratory. European players would follow afterward, with more tightly framed deployments and more cautious communication. Even so, the competitive pressure would be the same. If a U.S. SaaS vendor sharply reduces its costs thanks to agents and reinvests that margin into product, pricing, or customer acquisition, its French or European competitors will have to respond. The issue therefore concerns not only employment, but also the structural competitiveness of software companies in Europe.

Skilled employment, management, HR: what the ClickUp case says about the future of work

The main interest of this sequence is that it shifts the debate from theory to practice. For two years, economists, consulting firms, and researchers have published estimates on professions exposed to generative AI. Some studies by OpenAI, MIT, McKinsey, or the IMF have suggested that a significant share of office tasks could be automated or transformed. But those figures often remained abstract for employees. The ClickUp case gives this perspective material form: jobs disappear while agents take over operational functions.

The first lesson is that skilled employment is not protected in principle. A common mistake is to oppose automatable manual work and preserved intellectual work. Generative AI blurs that boundary. It does not easily replace judgment, complex negotiation, high-level original creation, or managerial responsibility. By contrast, it can absorb a large part of intermediate cognitive work: synthesizing, rephrasing, classifying, documenting, responding, tracking, following up, comparing, extracting, summarizing. Yet a significant share of service-sector jobs rests precisely on these operations.

The second lesson concerns management. If a company deploys thousands of agents, the role of managers evolves. They no longer oversee only people, but hybrid systems made up of humans, automations, and connected software. That requires new skills: defining quality indicators, managing exceptions, auditing agents’ decisions, documenting processes, arbitrating autonomy thresholds, organizing escalations, preventing drift. Tomorrow’s manager may spend less time coordinating a team of executors and more time orchestrating an informational production chain.

Human resources are also directly affected. The classic model of career progression in SaaS companies often relies on entry-level or mid-career roles: support, SDR, marketing operations, customer coordination, QA, documentation, junior analysts. These are precisely positions where AI can reduce needs. If these “entry steps” contract, the entire talent pyramid is affected. How do you train future managers if junior roles disappear or become scarcer? How do you transmit business knowledge if execution is entrusted to agents? The risk is creating organizations that are more efficient in the short term, but more fragile in terms of skills development.

For French-speaking audiences, this question is central. France and Europe have a significant base of skilled service-sector jobs in services, software, consulting, banking, insurance, telecoms, and administration. Many of these jobs include a significant share of documentary, relational, or procedural work. The ClickUp scenario suggests that AI will not first hit only the most visible creative or technical professions, but also the “invisible” functions that keep organizations running. These are often the jobs that absorb young graduates and structure career paths.

The psychological and cultural dimension must also be discussed. A company that replaces part of its workforce with agents sends an ambiguous message to those who remain. On one hand, it can present AI as a lever for refocusing on more strategic tasks. On the other, it installs the idea that any codifiable activity is potentially replaceable. That may encourage some employees to upskill in AI, but it can also generate anxiety, distrust, or lower engagement. In the long term, the success of these transformations will depend as much on human governance as on the technical performance of the models.

At the macroeconomic level, several scenarios coexist. The optimistic scenario holds that AI will eliminate certain tasks but create new jobs, new markets, and new layers of value, as other technological waves have done. The pessimistic scenario anticipates increased polarization of the labor market, with fewer intermediate positions and a concentration of value among the most strategic or most technical profiles. The ClickUp case does not settle this debate, but it provides an important clue: substitution can now be rapid enough to precede the creation of new roles. It is this temporal asymmetry that worries people.

In Europe, this tension could fuel an acceleration of regulatory debates. The European AI Act already frames certain categories of AI systems, but it does not directly answer all the questions linked to the restructuring of work. Future discussions could focus on transparency of internal uses, employee information, auditing agent performance, liability in case of error, or the traceability of automated decisions in HR, support, or customer relations functions. The issue is no longer only technological. It is becoming fully socio-economic.

What this implies for France and Europe, and what the coming months could confirm

For French companies, the ClickUp case acts as a strategic warning. Many executives have so far viewed generative AI as a tool for incremental improvement: saving time on writing, information retrieval, presentation preparation, meeting summaries. The logic of agents imposes a change of scale. It is no longer just about equipping employees, but about rethinking the processes themselves. Companies that remain at a purely experimental level risk finding themselves out of step with competitors capable of massively automating their operations.

This transition will, however, be neither uniform nor immediate. In France, several obstacles are slowing the generalization of agents: sensitivity around data hosted outside Europe, GDPR requirements, fragmentation of information systems, weakness of some internal document bases, lack of skills in agent orchestration, and caution from legal departments. But these obstacles are becoming investment projects. Large groups, scale-ups, and European software vendors are already working on safer architectures, on-premise or sovereign deployments, and more robust governance frameworks.

The French-speaking market could be affected first in three categories. First, SaaS vendors and digital companies, because they share with ClickUp a high density of informational processes. Second, support and customer relations functions, where return on investment is rapid. Finally, administrative and coordination roles in large service groups. In all these cases, the impact will not necessarily take the form of spectacular layoffs. It may come through unreplaced attrition, smaller teams, revised outsourcing, or a greater requirement for AI-assisted versatility.

Another likely effect concerns the hierarchy of skills. Profiles capable of designing, supervising, and integrating agents will gain in value. This includes automation specialists, data architects, AI product managers, security experts, analysts capable of formalizing processes, but also operational managers able to oversee hybrid systems. Conversely, profiles whose contribution is hard to distinguish from a standardized flow could face growing pressure. The labor market will not simply be “reduced” by AI: it will be resegmented.

For employees, the lesson is harsh but clear. Mastery of basic generative tools will probably not be enough. If everyone knows how to write a prompt, the advantage shifts to those who know how to structure a workflow, assess an agent’s quality, define guardrails, use business data, or transform a process. Future employability will depend less on occasional use of AI than on the ability to work with and around it, in environments where part of execution has already been automated.

The coming months will be decisive in determining whether ClickUp remains an isolated case or becomes a precedent cited across the industry. Several signals will need to be watched. First, how other SaaS vendors communicate about their own internal uses of AI. Next, changes in headcount in support, sales, and customer success functions. It will also be necessary to observe the metrics highlighted during financial results: reduction in service cost, improvement in operating margin, increase in revenue per employee, reduction in processing time. If these indicators improve in parallel with a rise in agents, the market will understand that substitution is no longer theoretical.

Customer reaction will also need to be monitored. A company can reduce its costs thanks to AI, but if service quality deteriorates, the advantage is fragile. Conversely, if agents enable better availability, faster responses, sufficient personalization, and lower prices, customers could validate this model. That is where the spread of the phenomenon will be decided. The lasting success of the “agentic” company will depend less on the announcement effect than on its ability to prove that large-scale automation genuinely improves the user experience.

Over the longer term, the ClickUp case could be reread as one of the first visible episodes of a deeper transformation of white-collar work. The digital office has already absorbed email, the cloud, real-time collaboration, videoconferencing, and workflow automation. AI agents add a new layer: that of quasi-autonomous software execution on tasks once carried out by skilled employees. For SaaS vendors, this could become a major competitive advantage. For European economies, it will be a test of adaptation speed. The companies that succeed will not simply be those that adopt AI earliest, but those that know how to redesign their organizations without destroying their capacity for learning, trust, and human innovation.