The Rundown

In this guide, you will learn how to set up AI teammates that are actually useful in ChatGPT's new Workspace Agents tool. The goal is one daily agent that handles a recurring morning task for you, instead of one more prompt you have to remember to run yourself.

Who This Is Useful For

• Operators and founders who do the same review, research, or reporting task every morning and want it off their plate
• Anyone managing a Notion database, lead pipeline, or shared folder who wants one agent to keep that system moving
• Teams already using ChatGPT Business who want something more durable than one-off prompts

What You Will Build

You will build a daily agent that owns one recurring workflow and runs it for you each morning.

In our demo, that agent reviewed a Notion database of published guides, looked for useful patterns, and generated three new guide ideas every day.

The finished agent gave us:

• a clear role and workflow
• connected apps and memory
• reusable starter actions
• a daily 9AM schedule
• one place to run, review, and refine the work over time

What You Need

• OpenAI says workspace agents are in research preview for Business, Enterprise, Edu, and Teachers plans
• One system you want the agent to own, like a Notion database, lead pipeline, or Google Docs folder
• Any integrations that workflow depends on, like Notion, Gmail, or Google Calendar

Pro tip: If you do not have ChatGPT Business yet, you can still use the same prompt and workflow design in another agent tool. The bigger lesson here is how to define ownership and recurring tasks clearly.

Step 1 Open the Agents Workspace and Start a New Agent

Go to chatgpt.com/agents. From there, click Create agent.

If you are on the wrong plan, ChatGPT will push you to sign up for a Business plan first first.

Once you are in, you will land on the new-agent screen with a single input box that asks what the agent should do.

Step 2 Give the Agent One Clear System to Own

Before you write anything, pick the system or process this agent should manage.

Good starting points:

• a lead pipeline
• a Google Docs folder
• a Notion database
• a daily report process

Then paste this prompt structure:

Create an agent that manages [lead pipeline, Google Docs folder, Notion database, etc.]. It should do 3 main tasks that can run autonomously on schedules to save me time.

In our test, we used an AI workflow researcher that owned a Notion database and looked for patterns in our published guides.

If you have a direct link to the database or tool you want it to manage, include that in the first setup message too. That gives the agent a cleaner starting point.

Pro tip: Point this at something annoying and sticky to maintain. A Notion database, Google Sheet, lead pipeline, task tracker, or reporting system is a much better fit than a vague brainstorming workflow.

Step 3 Connect the Right Apps and Let the Agent Scaffold Itself

After the first prompt, ChatGPT will walk you through the setup and ask for the tools it needs.

ChatGPT will fill in the rest of the details: role, workflow, apps, memory, and how the agent should behave on each run.

Pro tip: Keep approval on for anything sensitive. OpenAI says workspace agents can ask for permission before steps like editing a sheet, sending an email, or creating a calendar event. That is the right default.

Step 4 Add the Morning Schedule and Test It Once

Once the agent is shaped correctly, use the Schedule control in the upper right.

From there, add a recurring run. In our test, we set it to run every day at 9AM.

The finished agent gave us:

• a clear role and workflow
• connected apps and memory
• reusable starter actions
• a daily 9AM schedule
• one place to run, review, and refine the work over time

Pro tip: When you click on any of your agents you should see 3 pre-made tasks types for the agent to do in one click.

Going Further

If you don't want to start from scratch, click Browse Templates on https://chatgpt.com/agents.

OpenAI already has business-friendly templates for things like software review, finance close, sales, data analysis, and support. You can also click into a template and see which apps it expects and what kind of work it is meant to handle.

That makes setup pretty straightforward:

• copy a template that is close to your use case
• swap in your own integrations
• narrow it to one daily job
• let it run every morning