“Multiply what your team can create and accomplish.” That is how the video announcing one of OpenAI’s most anticipated new features ends: Workspace Agents. With this launch, the Sam Altman-led company is making it easier for teams to create shared agents that handle complex tasks and workflows, all within the permissions and controls set by their organization. In other words, agents can be shared, reused, and used to automate entire processes across teams.
This new feature is now available in preview for ChatGPT Business, Enterprise, Edu, and Teachers plans. In this first phase, agents will be free through May 6, 2026, after which they will be subject to pricing based on the credits used.
First, a necessary clarification 😉 Agents are systems that carry out a task and include a trigger, a process that may involve specialized skills, and tools or systems they can connect to.
At least, that is how OpenAI defines them.
In this context, OpenAI presents Workspace Agents as an evolution of GPTs, the applications users can create to perform specialized tasks in ChatGPT. These apps experienced a boom a couple of years ago, to the point that a “GPT store” was launched, much like the Play Store for mobile apps. In this regard, GPTs will remain available during this first phase, and OpenAI will later make it easier to convert GPTs into workspace agents.
The new agents are powered by Codex, and are designed to take on repetitive tasks in the workplace, from preparing reports to writing code and responding to messages. OpenAI made it clear from the outset that these new agents are specifically geared toward workplace use, so that teams “can create an agent once, use it together in ChatGPT or Slack, and improve it over time.” As you can see, with this move, OpenAI continues to move away from ChatGPT’s individual-use model and reinforces its positioning as a tool for automating entire business processes.
These agents have memory and, as expected, can improve as they are used. In that sense, OpenAI wants agents to become a practical way to “keep team knowledge up to date.”
As an example of how these agents can be used, the company explains that “our OpenAI sales team uses an agent to gather details from call notes and account research, qualify new leads, and draft follow-up emails directly in the salesperson’s inbox. It helps account teams spend less time gathering data and more time with customers.”
In any case, the most striking part of OpenAI’s approach is how much effort it has put into making these agents easy to create. In fact, you do not even need to think from the outset about how to configure one, since the company has made available a large template library for the most common use cases (each identified with a charming little googly-eyed icon).
The templates include common approaches such as “data analysis,” “customer service,” and “task management.”
If you prefer not to use one of those templates, you can build one “by hand,” although in reality ChatGPT does almost everything for you. You simply describe the task you want completed or attach a file. ChatGPT will help turn it into an agent by defining the steps and making it easier for you to connect the tools you need, add functions, and test it until it works exactly as expected.
In fact, simply by clicking the create agent button, the system kick off the entire creation process.
Of course, OpenAI is aware of the reservations that may exist around using agents in professional settings. That is why users will always be able to decide which tools and data an agent can use, what actions it can perform, and when it does or does not need approval. For more sensitive tasks, such as editing a spreadsheet, sending an email, or adding a calendar event, you can require the agent to ask for permission before proceeding.
Administrators will also be able to manage access, defining who has permission to use, create, and share agents.
Among the companies around the world that have already been able to use workspace agents is BBVA. In this regard, Elena Alfaro, Head of Global AI Adoption at BBVA, highlights that “at BBVA, we use thousands of GPTs every day, created by teams across the company for specific needs. Our early tests indicate that workspace agents will represent a major leap forward by managing end-to-end workflows, thanks to their ability to connect to data sources and execute tasks autonomously. At the same time, at BBVA we are working on a governance model that will allow us to coordinate and use these capabilities in a safe, responsible, and scalable way.”
As OpenAI explains in its announcement: “Teams perform at their best when information is easier to find, processes are simpler to follow, and people can get help without interrupting their workflow. Workspace agents are a first step toward that future: AI that works alongside people in the tools and conversations where work already happens, helping teams spend less time coordinating tasks and more time creating, developing, and making decisions that drive the business.”
In any case, this will not be OpenAI’s last step in this direction. The company has said it will continue rolling out new features in the coming weeks to help teams be more productive with less manual effort, including “enhanced dashboards to understand and optimize performance, more ways for agents to take actions across your business tools, and support for workspace agents in the Codex app.”
Image: ChatGPT Images 2.0
Your email address will not be published. Required fields are marked *
Δ