monday.com sells AI 'workers'

monday.com is not just adding another AI feature. It is trying to turn AI into labor you can shop for. In late March, the company launched Agentalent.ai, a managed marketplace where businesses can post a role, review qualified AI agents, and “hire” one for a defined job instead of buying a generic model or a loose bundle of automation tools (monday.com). That framing sounds cute on purpose. It is also the whole point. The product came out of monday agent labs, a new internal unit built to speed up enterprise adoption of what the company calls agentic work. monday.com says the marketplace was developed with AWS, Anthropic, and Wix, and that early users also include Mesh Payments and several implementation partners testing agents for marketing and operations work (monday.com). The company is pitching this as a familiar process for cautious buyers. You do not deploy “AI.” You define a role, test candidates, and onboard the one that performs. That matters because most companies still do not know how to buy agent software. A chatbot subscription is easy to understand. A model API is easy to meter. An autonomous system that can touch workflows, trigger actions, and make decisions is harder to slot into procurement, security review, and accountability. monday.com is trying to solve that by packaging agents as measurable outputs tied to operational roles, with authentication, authorization, and qualification checks before agents ever reach customers (monday.com). The launch did not come out of nowhere. In February 2025, monday.com said its AI strategy would center on “AI Blocks,” product power-ups, and a “Digital Workforce” of specialized agents that could work around the clock on things like project risk analysis and service issues (monday.com). By September 2025, it had rolled out a no-code agent builder and started describing its software less as a place to manage work than a place that “does the work for you” (monday.com). Agentalent.ai is what that strategy looks like once it leaves the slide deck. The company also spent March making its platform legible to nonhuman users. On March 11, monday.com announced new infrastructure for AI agents to sign up, authenticate, and operate directly inside the platform, with access to boards, automations, dashboards, and docs through the same account structure humans use (monday.com). That is the plumbing beneath the marketplace. You cannot “hire” an agent unless the software knows how to treat an agent as a first-class participant. This is also why the language of hiring is more than branding. Enterprise buyers want something closer to a governed service than a raw model. AWS has been pushing the same direction with its AI agents and tools catalog, which bundles prebuilt agents, tooling, and services for enterprise deployment (aws.amazon.com). Anthropic has been explicit about the problem too: agents are useful precisely because they are flexible and autonomous, which also makes them harder to evaluate with ordinary software tests (anthropic.com). monday.com is selling a workflow for that uncertainty. The company had already been nudging customers toward this model inside its existing marketplace. Its support docs describe “AI skills” for monday sidekick that let outside apps act inside the workspace, sometimes with permissions that cut across normal workspace-level admin restrictions (support.monday.com). Agentalent.ai takes that logic and formalizes it. The new thing is not just that an agent can act. It is that monday.com wants companies to treat that agent like a role with requirements, tests, access rules, contracts, billing, and a start date.