How to Build Workspace Agents in ChatGPT

TL;DR

Workspace agents are built by selecting Agents in the ChatGPT sidebar and describing an existing team workflow in natural language. These Codex-powered agents can access files, code, and tools, continue cloud-based work while a computer is closed, retain memory, and be shared through ChatGPT or Slack. The meeting-prep and software-review examples show how this approach turns multi-system processes into reusable team agents. Read on for the setup, use cases, and product distinctions.

Transcript

Hello and welcome back to build hours. I'm Victoria from product marketing and I'm here with Christina and Hojun. >> Hi, I'm Christina and I work on the engineering team for workspace agents. >> Hi, I'm Hojun. I'm on the solutions engineering team. >> Awesome. And today we're going to show you workspace agents in chatt. So if this is your first bui... Read More

Key Insights

• Workspace agents are codeex-powered tools in ChatGPT designed for complex, long-running tasks across multiple systems.
• They are built for teams, allowing shared work and running in the cloud even when your computer is closed.
• Agents can access files, code, and tools, and are designed to improve over time with memory capabilities.
• Creating an agent involves describing a workflow in natural language, which ChatGPT uses to build the agent.
• Agents can be customized and shared across teams, allowing for duplication and modification to suit specific workflows.
• Workspace agents can operate autonomously, running on schedules and interfacing with tools like Slack and Jira.
• The platform provides a centralized place to view all previous agent runs, ensuring transparency and auditability.
• Workspace agents are free during the research preview period, with credit-based pricing starting after May 6.

Questions & Answers

Q: How do you build a workspace agent in ChatGPT?

Click Agents in the ChatGPT sidebar, then describe a workflow your team already performs. ChatGPT turns that natural-language description into an agent, which you can build from a blank slate or by using a template.

Q: What are workspace agents in ChatGPT?

Workspace agents are Codex-powered agents built for complex, long-running work spanning multiple systems. They can access files, code, and tools, gather context, continue working in the background, and be shared with a team in ChatGPT or Slack.

Q: Which ChatGPT plans offer workspace agents during the research preview?

The research preview is available to ChatGPT Business, Enterprise, Edu, and ChatGPT for Teachers plans. The page states that workspace agents are free during the research preview, with credit-based pricing beginning after May 6.

Q: How does the meeting-prep workspace agent work?

Every morning, the agent checks the user's calendar and researches each customer on the web or in Google Drive. It then creates and links a meeting brief for every meeting, replacing hours of manual preparation and producing documents that can be shared with teammates.

Q: What does the software-review workspace agent do?

The software-review agent helps handle employee requests submitted in Slack. It routes the next steps according to the team's policies, demonstrating how an agent can coordinate a multi-system workflow.

Q: Can workspace agents be shared and customized by teams?

Yes. An agent can be shared in ChatGPT or Slack so an entire team can use it, and others can customize a shared agent for their specific workflows. Agents can also keep working in the cloud even when the user's computer is closed.

Q: How do workspace agents differ from Codex and the Agents SDK?

Workspace agents are designed for shared team work and cloud-based tasks that continue when a computer is closed. Codex is positioned for individuals working with a personal agent, while the Agents SDK is for teams building custom agents directly into products and customer experiences.

Q: What are some workspace-agent use cases?

Examples include preparing customer meeting briefs, reviewing software requests, turning a product brief into a website, supporting month-end close, and conducting vendor risk reviews. The vendor-review example researches sanctions exposure, financial health, and reputational risk before producing a structured report.

Summary & Key Takeaways

• Workspace agents in ChatGPT are designed to handle complex, long-running tasks across multiple systems, offering shared work solutions for teams. They run in the cloud, have memory capabilities for improvement over time, and can access files, code, and tools. Agents are built using natural language, allowing for easy customization and sharing across teams.

• Creating a workspace agent involves describing a workflow, which ChatGPT uses to build and configure the agent. They can be customized to suit specific workflows and shared with team members. Agents can operate autonomously, running on schedules and interfacing with tools like Slack and Jira to improve workflow efficiency.

• Workspace agents are currently free during the research preview period, with credit-based pricing starting after May 6. The platform provides a centralized place to view all previous agent runs, ensuring transparency and auditability. This makes workspace agents a powerful tool for improving team collaboration and productivity.