Create and manage agents using Agent Designer

Last updated: December 17, 2025
This page describes how to create, use, and manage an agent using Agent Designer in the Gemini Enterprise – Business edition.
 

Agent Designer overview 

Agent Designer is an interactive no-code, low-code platform for creating, managing, and launching single and multi-step agents in Gemini Enterprise – Business edition.

With Agent Designer, you can:

  • Create and launch custom agents using natural language prompts.
  • Visually edit agent workflows using the interactive Flow canvas.
  • Orchestrate complex tasks using multi-step agents (agents with sub-agents).
  • Connect your agents to various Google and external data sources and tools like Gmail, Google Drive, and Jira.
  • Share your agents to make them available across your team or organization.
  • Schedule agent executions to run tasks on a recurring basis.

Create a single-step agent

Before you begin: Set up any required connectors for your agent. For more information, see Connect your apps and third-party data.
 

You can create custom agents that automate tasks and/or integrate with your existing business applications. To create a single-step agent:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select  Agents.
  3. Select + Create agent.
  4. Select  Manage your data to choose the data sources you want your agent to use.
  5. Enter a prompt that describes your agent's purpose and desired behavior, then select Submit.

    For prompt guidance and examples, see Best practices for creating your agent.

    The system uses your prompt to generate an overview of your agent — including the name, description, instructions, and any specified connectors.
  6. Use Agent Designer to preview your agent, then select Create to publish your new agent.

Create a multi-step agent

Before you begin: Set up any required connectors for your agent. For more information, see Connect your apps and third-party data.
 

If you just created your new agent and are actively viewing the Agent Designer canvas, skip to Step 5.

To create a multi-step agent using the Agent Designer flow builder:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the navigation menu, select   Agents.
  3. Select + Create agent.
  4. Select Proceed to Builder.
  5. In the Agent Designer canvas, select the default main agent node.
  6. Configure the following details for your main agent in the details pane:
    Use conversational chat to improve your agent. You can use the chat box on the Agent Designer canvas to easily update your agent.
    Field Description
    Name Enter the name of your main agent.
    Model Select the Gemini LLM model that your main agent should use.
    Description Enter a summary of the purpose of the main agent.
    Instructions Enter a detailed set of instructions that define your agent's objectives, tasks, and behaviors.
    Data and tools Select (Add data sources & tools) to add data sources that your agent can use while generating responses or delegating tasks to subagents.
    Subagents  Select  (Add Subagents) to add and link subagents to the main agent.
  7. (Optional) Build your subagents. To build a subagent:
    1. Ensure you are in the Flow tab of the Agent Designer canvas.
    2. Select (Add subagent), and configure the details for each subagent.
  8. (Optional) Manage agent details. To manage agent information:
    1. Select the Details tab.
    2. Review and update the agent's name or description.
    3. In the Personalization section, add starter conversations.
    4. In the Knowledge section, select Add files to upload any additional reference materials for your agent.
      The listed data sources only include those which have been enabled for your account. Data sources can only be added or removed by a team administrator. See the Connect your apps article for guidance on configuring your data sources.
  9. Use the Preview tab to interact with your agent, and test and verify if your agent workflow is performing as expected.
  10. Select Create to publish your agent. If you exit the Agent Designer before publishing your agent, your agent will be set as a Draft.

Manage agent sharing

You can share and unshare your agents with other members of your organization. Shared agents appear in the From your organization section of the Agent gallery.

Note: Sharing is enabled by default in your Gemini Business instance. To learn how to enable or disable this feature, see Enable or disable agent sharing.

When an agent is shared, other users in your organization can:

  • Chat with the agent.
  • View the agent’s configuration (like instructions) in read-only mode. The read-only access helps users understand the agent's purpose and capabilities.
  • View and use all the data connectors associated with the agent.
Caution: When you share an agent, you are also sharing access to all its associated data sources and files. Users will be able to query the agent about the content of these sources, even if they do not have direct file or data source access.

For example, if you upload a confidential file to the agent’s knowledge section and then share the agent, any user with whom you shared it can ask the agent questions about that file's content.

Share an agent

To share an agent, do the following:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select  Agents.
  3. For the agent you want to share, select (Actions menu), and select  Share with org.

    The agent is now shared with your organization. A Shared icon appears on the agent card to indicate this.

Unshare an agent

To unshare an agent, do the following:

Note: Only the original owner can unshare an agent.

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select  Agents.
  3. For the agent you want to unshare, select (Actions menu), and select  Unshare with org.

    The agent is immediately removed from the From your organization section of the Agent gallery and is no longer accessible to other users.

Enable or disable agent sharing

Note: An administrator must configure agent sharing for your organization before users can share agents. For more information about the available roles in Gemini Business, see Available roles for your team.

To enable or disable the agent sharing feature in Agent Designer:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. Go to  Settings and click Admin settings.
  3. Select  Settings & preferences.
  4. In the Sharing section, toggle Enable agent sharing on or off as needed.

Use an agent

To use a published agent:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select Agents.
  3. Navigate to the Your Agents section and select the agent you want to use.
  4. Start interacting with the agent using natural language prompts to perform your desired tasks.
    While interacting with the agent, you can also select (Data) to directly edit the available data connectors.

Depending on your agent plan, task configuration, and connector setup, you can use an agent to assist you in a variety of tasks.

Task type Description
Answer questions Agents can provide responses to a wide range of questions by accessing and synthesizing information from connected data sources.
Search and discovery Agents can search across multiple connected data sources.
Content generation Agents can help generate content, summarize documents, and create reports.
Task automation Agents can automate multi-step processes.
Data analysis Agents can analyze data to provide insights.

Schedule agent executions

You can configure your agent to execute tasks automatically on a recurring basis. A schedule serves as an automated trigger for the agent's predefined instructions or prompts.

Note: Scheduled agents run automatically for personal tasks, such as creating your own calendar invites or emailing yourself. However, any action involving other people will pause for your review and approval.

Add an execution schedule

To add an execution schedule for your agent, do the following:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select  Agents.
  3. For the agent you want to edit, select (Actions menu), and select Edit.
  4. In the Agent Designer canvas, click the Schedule tab.
  5. Click + Add schedule and configure the recurrence and execution time:
    • Repeat frequency: select how often the agent should run to execute the provided prompt. The following options are available:
      • Hourly: runs the task every specified number of hours and at a specific minute within that hour.
      • Daily: runs the task once per day at the specified time.
      • Weekly: runs the task once per week on the specified day and time.
      • Monthly: runs the task once per month on the specified day of the month and time.
      • Annually: runs the task once per year on the specified month, day, and time.
    • Execution time and timezone: define the specific interval settings based on your selected frequency and select the execution timezone.
    • Prompt: enter the specific prompt (or instruction) the agent will receive at the scheduled time. This defines the task the agent will execute.
  6. Click Save to create the schedule and add it to you agent in a Draft state.
  7. Optional: Test your schedule to verify if it functions as expected.
  8. To activate your schedule and relaunch your agent with the updated changes, click Update.

Test the execution schedule

After configuring a schedule, you can immediately test the agent's execution to ensure it performs the task as expected without waiting for the next scheduled interval.

To test the execution schedule, do the following:

  1. In the Agent Designer canvas, click the Schedule tab.
  2. Find the schedule card that you want to test and click Run in preview.

    This triggers the agent using the saved prompt and configuration.

Edit an execution schedule

Edit an execution schedule to make changes to the frequency, time, timezone, or prompt.

To edit an existing execution scheduled, do the following:

  1. In the Agent Designer canvas, click the Schedule tab.
  2. Find the schedule card that you want to test and click Edit.
  3. Modify the Repeat Frequency, Run Time, Timezone, and the Prompt as desired.
  4. Click Save.
  5. Click Update to ensure all changes, including the new schedule, are saved and applied to the agent.

Delete an execution schedule

If a recurring task is no longer needed, you can delete and remove it from the Agent Schedules list.

To delete an execution schedule, do the following:

  1. In the Agent Designer canvas, click the Schedule tab.
  2. Find the schedule card that you want to test and click Delete.
  3. Confirm the deletion to permanently remove the execution schedule for your agent.

Refresh agent schedules

Agent schedules operate using your user credentials. To ensure data privacy and security, these credentials expire every 14 days. You must manually refresh your agent schedules to ensure they continue to run automatically without interruption.

To refresh schedules for all agents, do the following:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select Agents.
  3. Go to the Your agents section and click Refresh the existing schedules.

    Your user credentials are refreshed, and all scheduled agents will continue to run for another 14-day period.

Limitations for agent schedules

  • You must manually refresh your schedules every 14 days to prevent all your scheduled agents from failing to execute.
  • There can be a maximum delay of 5 minutes for a scheduled agent to start running.
  • Schedules are only supported in multi-region locations.

Edit an existing agent

To edit your agent while you're interacting with it, select the Open in Agent Designer button in the top right corner. If you are actively viewing the Agent Designer canvas of the agent you wish to edit, skip to Step 4.

To edit your agent:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select  Agents.
  3. For the agent you want to edit, select (Actions menu), and select Edit.
  4. Use Agent Designer to configure the agent as desired.
    Use conversational chat to improve your agent. You can use the chat box on the Agent Designer canvas to easily update your agent.
    Field Description
    Name Enter the name of your main agent.
    Model Select the Gemini LLM model that your main agent should use.
    Description Enter a summary of the purpose of the main agent.
    Instructions Enter a detailed set of instructions that define your agent's objectives, tasks, and behaviors.
    Data and tools Select (Add data sources & tools) to add data sources that your agent can use while generating responses or delegating tasks to subagents.
    Subagents  Select  (Add Subagents) to add and link subagents to the main agent.
  5. Select Update to save and relaunch the agent.

Delete an existing agent

To delete an agent:

  1. Sign in to Gemini Enterprise – Business edition (business.gemini.google/).
  2. In the app navigation panel (≡), select Agents.
  3. For the agent you want to delete, select (Actions) and select Delete.
  4. Select Delete to permanently delete the agent.
    Upon confirmation, your agent is deleted and removed from your Gemini Enterprise – Business edition account.

 

Best practices for creating your agent

When you create a new agent, the prompt you write helps Gemini Enterprise – Business edition understand your goals and generate the best possible agent plan.

The following are a few guidelines and best practices to help you get clear and relevant responses:

Goal Description Examples
Provide enough context to eliminate ambiguity
  • Be specific about what you want your agent to do.
  • Specify the level of detail, the types of examples to use, or the required output.
  • Additional context helps generate an agent plan that is more relevant and specific to your needs.

 Recommended: “Create an agent that drafts follow-up emails for new sales leads from my CRM. It should search for the contact's company, summarize their business, and suggest how our product can help them.”

 Not recommended: “Create an agent for sales emails.”

Set boundaries or limitations on the agent’s output
  • Define what the agent should and should not do.
  • You can set limits on the length of its responses, the format, the tone, or specific elements to include or exclude.
  • Setting boundaries helps the agent stay focused and produce more predictable results.

Recommended: “Create an agent that drafts follow-up emails for sales leads. The drafts should always use a professional tone, be no more than three paragraphs, and never include personal information about the client.”

Not recommended: “Create a sales email agent.”

Prompt the chatbot to ask you questions
  • Instruct the chatbot to ask you questions.
  • Forcing the chatbot to ask you questions helps gather details the chatbot might have otherwise missed.

Recommended: “I need an agent that can help me draft sales emails. First, ask me what information you need to effectively qualify a lead and draft a personalized message.”

Not recommended: “Create an agent to handle sales emails.”

Example of an effective prompt  “Create an agent that automatically drafts sales emails to leads. Use the Salesforce connector to access my leads list. First, ask me to confirm which leads you should prioritize and what the main selling points are for each. Once you have this information, your drafts should be no more than three paragraphs and use a professional tone. Do not include any technical jargon or slang and never include any personal information about the client."
Main menu
7682235135483321540
true
Search Help Center
false
true
true
true
false
false
false
false