You can use Google Workspace Studio only if your organization supports it. To use generative AI features, your Google Account must be allowed to use Gemini.
If you're signed in with a school account and are under 18, you can’t use any AI features in Workspace Studio. If an agent containing AI steps is shared with you, those steps will be removed when you open the agent.
Agent building blocks
In Workspace Studio, you automate your everyday, routine tasks across Google Workspace services by creating agents. Agents start based on an event or schedule and then do a set of actions automatically. An agent has 3 main parts:
- Starter: This is the event or schedule that starts your agent. For example, when you get an email message from specific people, or every Monday at 9 am.
- Steps: These are the actions that the agent does after it starts. Most agents have several steps. For example, ask Gemini to extract action items, draft a reply, or add content to a document.
- Variables: Each step has variables, which are dynamic placeholders for information that you can use in later steps. For example, Gemini’s response to your prompt, the sender of the message that started the agent, a form response, and more. Learn more about variables in the Workspace Studio Help Center.
Create an agent with AI
You can use AI to generate an agent in Workspace Studio. Just enter what you want the agent to do and when. Gemini puts together the right starter and steps, and drafts any content. Then you can edit before turning it on.
- Go to https://studio.workspace.google.com.
- Click Describe tasks to get started and enter a description in English of what you want your agent to do and when.
For best results, provide these details:
- How or when to start the agent. For example, enter “every Monday at 4:30” instead of “each week.”
- The names of the apps to use. For example, enter "notify me in Chat" instead of "notify me."
- The specific people. For example, enter your manager's email address instead of "my manager."
Learn more tips to create agents with AI.
- Click Create with AI. Workspace Studio builds a custom agent for you.
- Click and review each step to make sure it does what you want. You can edit, delete, reorder, and add more steps. Tips:
- To use information from previous steps, add variables by clicking
Variables. Learn more about using variables.
- To do some steps only when certain conditions are met, use conditional steps.
- If you don't see a step that does what you want to do, you can install a custom step from Google Workspace Marketplace or build your own. Learn more about custom steps.
- To use information from previous steps, add variables by clicking
- If your agent has an AI-powered step, such as Ask Gemini or Summarize, review which sources Gemini can use. By default, Gemini can reference all content that you can access in Google Workspace and connected integrations. To prevent unintentional sharing of your organization's data by the agent, choose Web only. Learn more about Gemini sources.
- If your agent has a third-party integration step, click the step and then Connect. Follow the instructions to connect the integration and your Google Account. Learn more about integration steps.
- (Optional) To try your agent, click Test run. A test run runs your agent once, taking real actions so you can see the outcome. Select the starting conditions, then click Start. Learn more about test runs.
- Click Turn on.
Start with a template
A template is an agent already built for a common task. Most require some input from you to get the details right. You can add or remove steps to meet your needs.
- Go to https://studio.workspace.google.com and click Discover.
Tip: You can also create agents from a template in the Workspace Studio side panel in Gmail, Drive, or Chat. In the app, at the top click
Studio.
- Find a template you want to use and click it.
- Customize the first step of your agent (the starter).
- Click the next step and customize it.
- To use information from previous steps, add variables by clicking
- If your agent has an AI-powered step, such as Ask Gemini or Summarize, review which sources Gemini can use. By default, Gemini can reference all content that you can access in Google Workspace and connected integrations. To prevent unintentional sharing of your organization's data by the agent, choose Web only. Learn more about Gemini sources.
- If your agent has a third-party integration step, click the step and then Connect. Follow the instructions to connect the integration and your Google Account. Learn more about integration steps.
- To use information from previous steps, add variables by clicking
- (Optional) Add more steps to customize your agent. In the left panel, click
Menu and then the arrow for the direction you want to move it.
- If you don't see a step that does what you want to do, you can install a custom step from Google Workspace Marketplace or build your own. Learn more about custom steps.
- (Optional) To try your agent, click Test run. A test run runs your agent once, taking real actions so you can see the outcome. Select the starting conditions, then click Start. Learn more about test runs.
- Click Turn on.
Create a custom agent manually
To build an agent from scratch, you can select your own starter and steps.
- Go to https://studio.workspace.google.com and click
- Select how your agent starts.
- Set up the starter.
- At the left, click
- Select the first step your agent will do.
- Learn tips to use AI steps in agents.
- To do some steps only when certain conditions are met, use conditional steps.
- If you don't see a step that does what you want to do, you can install a custom step from Google Workspace Marketplace or build your own. Learn more about custom steps.
- Set up the step. In most boxes, you can use information from a previous step, such as Gemini’s response or the name of the person who sent the message, by adding a variable.
- To use information from previous steps, add variables by clicking
- If your agent has an AI-powered step, such as Ask Gemini or Summarize, review which sources Gemini can use. By default, Gemini can reference all content that you can access in Google Workspace and connected integrations. To prevent unintentional sharing of your organization's data by the agent, choose Web only. Learn more about Gemini sources.
- If your agent has a third-party integration step, click the step and then Connect. Follow the instructions to connect the integration and your Google Account. Learn more about integration steps.
- To use information from previous steps, add variables by clicking
- Add more steps and set up as needed.
- (Optional) To try your agent, click Test run. A test run runs your agent once, taking real actions so you can see the outcome. Select the starting conditions, then click Start. Learn more about test runs.
- When your steps are all set up, click Turn on.