OpenAI Launches Cloud-Based AI Agents for Team Collaboration

OpenAI introduces 'workspace agents' in ChatGPT for team collaboration.

OpenAI has introduced ‘workspace agents’ in ChatGPT. Teams can create shared assistants to tackle complex tasks and lengthy processes.

These virtual assistants are powered by Codex. They are capable of handling tasks typically performed by humans—from preparing reports and writing code to responding to messages.

AI agents operate in the cloud, enabling them to execute assignments even in the absence of a human user. They are designed for collaborative use within an organization. Teams can create an assistant once and then jointly use it in ChatGPT or Slack.

“AI has already helped individuals become more productive in their personal work. However, many key processes within a company rely on shared context, task handovers between employees, and inter-team decisions. Workspace agents are designed for such work: they can gather context from necessary systems, follow team processes, request approval when needed, and coordinate work across different tools,” the announcement states.

As an example, OpenAI highlighted an agent used by a sales team. It can consolidate information from call notes and client research, qualify new leads, and draft follow-up emails for the sales manager.

“This allows more time to be spent on client interactions rather than ‘piecing together’ details,” the company noted.

‘Workspace agents’ are available under research preview access in ChatGPT Business, Enterprise, and Edu plans, as well as in ChatGPT for Teachers (K–12).

Privacy Safeguarded by OpenAI

Separately, OpenAI has released a free Privacy Filter tool, which cleanses confidential information in conversations with ChatGPT before it is seen by the chatbot. This includes tax returns, medical records, work emails with client names, or API keys.

The Privacy Filter is available on Hugging Face and GitHub and comprises 1.5 billion parameters. Anyone can download the software and use or modify it as needed.

When a user inputs text in a conversation with the chatbot, the program automatically replaces all confidential data with generic information like [PRIVATE_PERSON] or [ACCOUNT_NUMBER].

The filter scans text for eight categories of personal information: names, addresses, emails, phone numbers, URLs, dates, accounts, as well as sensitive data like passwords and API keys.

OpenAI reported that the model is quite effective at recognizing important information, scoring 96% in a standard test using the PII-Masking-300k dataset.

Back in April, Sam Altman’s startup released the ‘thinking’ image generator ChatGPT Images 2.0—an “advanced model capable of solving complex visual tasks and creating precise, ready-to-use works.”