Create an Agent in Procore

AI agents have specialized instructions designed to handle specific project tasks with more precision. You can give the AI the context and tools needed for your specific workflow. For example: An 'RFI Validator Agent' can make sure the AI stops acting as a generalist and views the data through the lens of RFI validation. Learn how to add the following agent types:

• Pre-built agents like our Punch List Management Agent.

• Partner-built agents like an agent that recommends equipment rentals that align with your project scope.

• Your own agents like one that drafts weekly owner reports using your company’s template and tone.

• Prerequisites

  • Datagrid Enterprise or Procore AI's Digital Coworker Enterprise.

  • To create or add an agent in Procore AI: A Company Admin must enable the project and assign you a role.

• Duplication for Procore AI customers: You can create an agent in Datagrid that then also shows up in the Procore AI side panel, and vis versa.

• Inheritance: Most agents start by inheriting settings from a master template (like Deep Search).

• Selection: Users can select more than one agent at a time to combine their capabilities for a task. To learn more about the controls can affect how your agent is used, see Manage Your AI Chat Controls.

• Validation: AI can make mistakes. Always check all AI work, results, and answers against your own records to ensure accuracy.

1. ClickAI on the top right to open the side panel.

2. Click the agent name at the bottom of the chat field. If it's in auto mode, click Procore AI.

3. Click + Create an Agent.

4. Choose one of the agent creation options below.

Add a Pre-Built Agent

Add one of our 50+ pre-built agents so that you can customize it or use it as-is.

1. Click + Create Agent.

2. Browse the categories in the 'Find Your Perfect AI Agent' window (e.g., Construction, Data Analysis, or Jobsite Management).

3. Click on an agent, such as the RFI Validator Agent or Submittal-Builder Agent to see example prompts that people use with that agent and decide if you want to create an agent using this template.

4. Click to select the connector(s) you want added to the agent. Whatever you connect gives the agent access to that data when making recommendations or providing information to you.

   • Recommendation: Add Procore as a connector so the agent's responses reflect the data in Procore.

5. Click Create.
   The agent is added to your list and automatically selected for your current chat session. The agent's configuration panel also opens, where you can customize it further.

6. To begin testing your new agent's output, type a question or request into the prompt composer and click thesend arrow or press ENTER on your keyboard.

   • Learn more about using Procore AI, including how to review responses.

Create a Custom Agent

If the available templates don't match your needs, you can create your own with your own prompt.

1. Click + Create Agent.

2. In the 'Find Your Perfect AI Agent' window, click the Custom Agent button.

3. Click the button that aligns with the goal you have for the agent and the text field will populate with a starter prompt that aligns with that goal. Edit or add to the prompt until you are satisfied.

   • Or, click Start with General Assistant to create a more generic agent that is pre-loaded with capabilities designed for Procore users, such as access to most Procore tool datasets and a few common AI tools such as: web search, semantic search, file finding, SQL query searching, and CSV file analysis and restructuring.

4. Click Generate.
   The agent is added to your list and automatically selected for your current chat session. The agent's configuration panel also opens, where you can customize it further.

5. To begin testing your new agent's output, type a question or request into the prompt composer and click thesend arrow or press ENTER on your keyboard.

   • Learn more about using Procore AI, including how to verify and audit responses.