Run agents using workflows

Use the Run Agent workflow action to deploy AI-powered agents directly from workflows. Learn more about using agents and assistants in Breeze Studio. You can configure which agent to run, what information it receives, and the output generated. Then, use the generated outputs to update records, automate follow-ups, and more. 

Please note: you can manage AI feature access in your AI settings and configure what data is shared. Review HubSpot's AI Trust FAQs and AI Model Cards for detailed information on AI security controls, data use, and compliance.

Currently, the Run agent action does not require HubSpot credits to execute. However, the action will consume credits when the feature is no longer in BETA. 

Use the Run Agent action 

1. In your HubSpot account, navigate to Automation > Workflows.
2. Click the name of an existing workflow or create a new workflow.
3. Set your enrollment triggers.
4. Click the + plus icon to add an action.
5. In the left panel, click to expand the AI section. Then, click Run agent. 
6. Click the Agent to run dropdown menu and select an agent. Only existing agents that have already been added to the account will appear when selecting an agent. Learn how to install agents. 
7. Depending on the agent selected, manually enter or add your input data using the data panel. For example, if you're using the Company Research Agent, you'll need to add the Domain or Company Name for the company. 
    

. 

8. In the Outputs section, click the Agent response format dropdown menu and select from the following: 
   • Text response: this will return the full agent's response in text format. 
   • Structured data: this will return the specific defined outputs. For example, if you're using the data research agent, you may want a specific pieces of information, such as the company's size or industry, or even an executive summary including recent news about the company.
     • To manually set your outputs, click + Add an output. Then, configure your outputs:
       • Output name: the name of the output returned. This will appear when selecting the action output for use with other workflow action types. 
       • What data should be returned: use a prompt to determine what kind of data should be returned. For example, if you're trying to determine the company's size, you can use the prompt: the number of employees in the company. Then, select number for the following data format field.
       • Data format: you can select from the following data types: text, number, boolean, date, datetime, enumeration, and phone number. 
     • To use AI to generate outputs, click Generate from instructions. Pre-filled structured outputs will be automatically generated by AI, based on the selected agent's instructions.
9. At the top, click Save. 
   
   

Use outputs with other workflow actions

After adding the Run agent action to your workflow, you can use the generated output in other workflow actions where the data panel is used. 

1. After setting up your action, in the same workflow, click the + plus icon to add an action.
2. Select a workflow action.
3. In the All data tokens panel, in the Action data section, click Run agent.
4. If you've chosen to use Text response when setting up your run agent action, select Text response.
5. If you've chosen to use Structured data when setting up your run agent action, select a specified output. 
6. Continue setting up your workflow action. 
    

Example use cases

The following examples outline the different ways you can use the Run agents action in workflows to automate more complex processes:

• Simplify research for sales reps: use the action with the Company research agent to review an executive summary of a contact's company before setting up a first meeting or calling the contact. You can also copy this data to a custom property on the record. 
• Accurately nurture leads: use the Customer health agent to categorize a record, then send them down different workflow branches.