OpenAI workspace agents raise the stakes for monday.com teams

A new kind of competition is taking shape

OpenAI just moved ChatGPT closer to the work orchestration layer monday.com has spent years building. Workspace agents are built for shared, repeatable workflows, not one-off prompts, and that changes the comparison from “AI helper” to “who owns the work.” For monday.com teams, that is the sharper story: if agents can already route feedback, pull metrics, and kick off approvals inside tools people use every day, customers will start asking why those handoffs should live anywhere else.

The stakes are especially high because OpenAI is not pitching a toy feature. Workspace agents are available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans, and OpenAI says they are Codex-powered, designed to handle complex tasks and long-running workflows. That puts the product squarely in the territory where project management, sales ops, and internal tooling overlap.

What OpenAI is offering

OpenAI’s workspace agents are built to be shared, scheduled, and monitored, which makes them much more operational than a standard chatbot. Teams can create agents from templates or from scratch, connect them to Google Drive, Google Calendar, Slack, and Microsoft SharePoint, and add skills, files, and custom MCP servers. They can also share agents privately, by link, or through a workspace directory, then track version history and analytics as those agents evolve.

The examples OpenAI chose are telling. It is not leading with marketing copy or casual productivity tricks. It is pointing to software review, product feedback routing, weekly metrics reporting, lead outreach, and third-party risk management, all tasks that depend on context, approvals, and a consistent chain of action. It is also saying these agents can run in the cloud, so work can keep moving even when no one is actively watching the screen, and they can be used in ChatGPT or in Slack channels.

That matters because the launch is built around governance as much as capability. Workspace admins can control who can build, publish, and use agents in Slack. Eligible Enterprise workspaces get the feature off by default at launch, and Enterprise tenants using EKM are excluded for now. In other words, OpenAI is not just courting users with convenience; it is trying to satisfy the compliance and control demands that decide enterprise deals.

Why this hits monday.com directly

For monday.com, the competitive implication is straightforward: buyers will increasingly compare monday’s automation and orchestration story with an AI layer that sits inside the chat and document tools employees already use all day. That raises the bar for integration depth, permissions-aware actions, and measurable workflow value. A polished draft is no longer enough. Customers will want proof that AI can trigger real work, respect access boundaries, and show what happened afterward.

That is where monday.com’s own response becomes important. On March 11, 2026, the company announced infrastructure that lets AI agents sign up, authenticate, and operate directly within the monday.com platform. Roy Mann said monday.com is building the infrastructure that allows humans and AI agents to collaborate directly, and the company says those agents can organize projects, update workflows, trigger automations, generate reports, and coordinate work across teams. That is the right strategic direction, but OpenAI’s move shows that monday.com is now defending that narrative against a much larger AI layer that already lives inside the tools many teams open first.

There is also a practical platform question underneath all of this: where does the work actually happen? monday.com has documented a ChatGPT integration path through monday MCP, including a native read-only connector and a fuller MCP setup for broader functionality. That means ChatGPT is not a distant threat on the horizon. It is already part of the environment monday.com has to work with, and now OpenAI is trying to make that environment more capable.

What product teams need to watch

If you work in product or engineering at monday.com, the issue is not whether OpenAI can write a better prompt. It is whether monday.com can keep proving that its own Work OS primitives are the place where work becomes structured, assigned, and measurable. OpenAI’s launch suggests the market is moving toward agentic systems that can schedule, hand off, and document work without constant supervision. That pushes monday.com to show more than surface-level AI assistance.

The clearest product pressure points are:

• Permissions-aware execution, not just content generation.
• Audit trails and version history that customers can trust.
• Workflow actions that span tools instead of stopping at summaries.
• Admin controls that satisfy IT and security teams.
• Reporting that shows completed work, not just generated text.

OpenAI’s own feature list underlines those expectations. If users can run recurring jobs, share agents through a workspace directory, and see analytics on what those agents do, then monday.com has to make its own AI story feel equally operational. The more AI can act like a teammate, the more the platform has to behave like a system of record for that teammate’s work.

What sales teams will hear from buyers

For sales, the conversation is going to shift fast from novelty demos to control, ownership, and integration. OpenAI’s examples already sound like the kinds of business cases enterprise buyers will bring to a monday.com rep: account research, lead outreach, approval workflows, recurring reporting, and risk checks. Once those examples move from hypothetical to available in ChatGPT and Slack, “we also have AI” stops being persuasive.

That also means sales conversations will need to get more specific about where monday.com wins. The strongest answer is not that monday.com has AI, but that it has an operating layer for teams that need to see ownership, dependencies, and status in one place. If OpenAI can draft the work and route it through chat, monday.com has to show that its platform is where the work is structured, tracked, and governed end to end.

The investor shadow hanging over the story

This is not landing in a vacuum. CNBC reported on February 9, 2026, that monday.com shares had already lost about half their value in 2026 and more than three quarters from their November 2021 high, as AI disruption fears weighed on the stock. That gives every OpenAI product move extra weight. Investors are not just watching feature parity; they are looking for evidence that monday.com can defend its category as AI changes how teams coordinate work.

That is the real meaning of workspace agents for monday.com. OpenAI is not trying to replace project management software with chat. It is trying to own the layer where work gets assigned, tracked, and pushed forward. For monday.com, the response has to be equally direct: prove that the Work OS remains the place where agents, people, permissions, and outcomes come together, or risk watching workflow itself drift toward the chat tools teams already trust.