OpenAI Launches Workspace Agents in ChatGPT for Teams

OpenAI has introduced workspace agents in ChatGPT, letting teams create shared AI agents that handle complex tasks and long-running workflows while operating within the permissions and controls set by their organization. The feature is available now in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans, and it will be free until May 6, 2026, when credit-based pricing begins.

The stakes are considerable. OpenAI is betting that the next phase of enterprise AI is not individual productivity but shared, organizational work — the handoffs, approvals, and cross-team processes that define how companies actually run. Workspace agents are positioned as the evolution of GPTs, the customizable chatbots OpenAI launched in late 2023, but with a fundamentally different operating model: they run in the cloud, keep working when the user logs off, and are designed to be built once and shared across an entire organization in ChatGPT or Slack.

What workspace agents do

According to OpenAI, workspace agents are powered by Codex in the cloud, which gives them access to a workspace containing files, code, tools, and memory. The company says the agents go beyond answering prompts: they can write or run code, use connected apps, remember what they have learned, and continue work across multiple steps.

"AI has already helped people work faster on their own, but many of the most important workflows inside an organization depend on shared context, handoffs, and decisions across teams," OpenAI wrote in its announcement. "Workspace agents are designed for that kind of work: they can gather context from the right systems, follow team processes, ask for approval when needed, and keep work moving across tools."

The company points to its own sales team as an early case study. An agent there pulls together details from call notes and account research, qualifies new leads, and drafts follow-up emails directly in a representative's inbox. OpenAI says the result is that account teams spend less time stitching together details and more time with customers.

Teams can schedule agents to run at set times, or deploy them in Slack so they pick up requests as they arrive. OpenAI's product team, for instance, built an agent that proactively answers employee questions in Slack channels, links relevant documentation, and files a ticket when it identifies a new issue. Today, teams can interact with agents in ChatGPT and Slack, with additional surfaces promised soon.

Five example agents

OpenAI published five agents its internal teams have built, framed as templates other organizations can replicate:

• Software Reviewer: Reviews employee software requests, checks them against approved tools and policies, recommends next steps, and files IT tickets when needed.
• Product Feedback Router: Monitors Slack, support channels, and public forums, then turns feedback into prioritized tickets and weekly product summaries.
• Weekly Metrics Reporter: Pulls data every Friday, creates charts, writes the summary, and shares a report with the team.
• Lead Outreach Agent: Researches inbound leads, scores them against a qualification rubric, drafts personalized follow-up emails, and updates the CRM.
• Third-Party Risk Manager: Researches vendors, assesses signals like sanctions exposure, financial health, and reputational risk, and produces a structured report.

Templates for finance, sales, marketing, and other functions are also available, each with built-in skills and suggested tools. Creating an agent requires no code: a user clicks Agents in the ChatGPT sidebar, describes a recurring workflow or drops in a file, and ChatGPT guides them through defining steps, connecting tools, adding skills, and testing the result.

The accounting angle is one of the more concrete demonstrations. OpenAI's accounting team built an agent that prepares key parts of month-end close — journal entries, balance sheet reconciliations, variance analysis. The company says it completes the work in minutes, generates workpapers with underlying inputs and control totals for review, and follows internal policies. Because agents have memory and can be guided and corrected in conversation, OpenAI argues they improve over time and become a practical way to keep team knowledge current: build once, improve through use, then share or duplicate for new workflows.

Governance and safeguards

For enterprises, the governance story may matter as much as the capabilities. When a user delegates work to an agent, they decide what tools and data it can use, what actions it can take, and when it needs approval. For sensitive steps — editing a spreadsheet, sending an email, adding a calendar event — the agent can be required to ask permission before proceeding.

ChatGPT Enterprise and Edu admins can control which connected tools and actions user groups can access, and they can manage who has permission to use, build, and share agents. OpenAI says built-in safeguards help agents stay aligned with user instructions when they encounter misleading external content, including prompt injection attacks — a persistent security concern for agentic systems that read untrusted data from the web, inboxes, and chat channels.

A Compliance API gives admins visibility into every agent's configuration, updates, and runs, and allows them to suspend agents if needed. After an agent is shared, analytics show how many runs it has completed and how many people are using it. OpenAI says admins will soon be able to view every agent built across their organization in the admin console, including usage patterns and connected data sources.

Early customer feedback

OpenAI published one early testimonial from Rippling, the HR and payroll software company. "The hard part of building an agent is not the model. It's the integrations, memory, the user experience," said Ankur Bhatt of AI Engineering at Rippling. "Workspace agents collapsed that work, so one of our Sales Consultants built, evaluated, and iterated a Sales Opportunity agent end to end without an engineering team. It researches accounts, summarizes Gong calls, and posts deal briefs directly into the team's Slack room. What used to take reps 5-6 hours a week now runs automatically in the background on every deal."

That 5-6 hours per week figure is the kind of concrete claim enterprise buyers will scrutinize as the product moves from free preview to paid, credit-based pricing next year.

The GPTs question

Existing GPTs will remain available while teams test workspace agents with their workflows, OpenAI confirmed in an editor's note. The company says it will soon make it easy to convert GPTs into workspace agents — a signal that GPTs are effectively being deprecated as the product line matures into something more operationally capable.

What comes next

OpenAI says additions in the coming weeks will include new triggers that start work automatically, better dashboards for understanding and improving performance, more ways for agents to take action across business tools, and support for workspace agents in the Codex app.

The competitive context is hard to ignore. Enterprise agentic AI has become the central battleground among major vendors, and OpenAI is distinguishing its approach by anchoring agents inside the tools where work already happens — ChatGPT today, Slack as a first-class surface, with more to come. "Teams do their best work when knowledge is easier to find, processes are easier to follow, and people can get help in the flow of work," the company wrote. Whether shared, continuously running agents become standard enterprise infrastructure or another layer of workflow complexity will depend largely on what the credit-based pricing looks like when the free period ends on May 6, 2026.