A Background Agent is a task-oriented AI agent that users can manually run or automate through Smart Flow. It can help users review CRM records, analyze data, create follow-up tasks, and perform scheduled or triggered tasks.
Topics Covered:
Create a Background Agent
To create a Background Agent, follow the steps below:
- Navigate to AI Studio > Team Agents > Browse Agents.
- Click Create From Scratch.
- Select Build a Background Agent.
When to Use a Background Agent
A Background Agent is useful when a task needs to be performed against CRM data manually or automatically.
Some common use cases include:
- Reviewing stale deals.
- Analyzing closed-lost deals.
- Summarizing calls.
- Identifying accounts at risk.
- Creating follow-up tasks.
- Sending scheduled reports.
Example scenarios:
A sales manager can configure a Background Agent to review stale deals every morning and create follow-up tasks for deals that need attention.
A team can use a Background Agent to summarize calls from the previous week and identify accounts that may be at risk.
Configure the Background Agent
After selecting Build a Background Agent, configure the agent using the Agent Builder.
- Agent Avatar: Choose an avatar for the Background Agent. The avatar provides the visual identity of the agent across AI Studio and the agent experience.
- Agent Name: Enter the name of the Background Agent. The agent name is required.
- Short Description: Add a short description explaining what the Background Agent does.
-
Agent Instructions: Enter instructions that define what the Background Agent should handle, what it should check before making a decision, and what result it should provide.
-
For example:
- Review open deals with no activity in the last 14 days.
- Explain why each deal needs attention.
- Recommend the next best action.
- Create a task only when the action policy allows it.
- Do not guess missing information.
-
For example:
-
Add Skills: Agent Skills are optional reusable capabilities that can help you build the Background Agent faster. Click Add Skill to add a skill. To learn more about Agent Skills, see Skills.
-
Examples include:
- Deal Health Analysis
- Lead Qualification
- Account Risk Review
- Follow-up Email Writer
- Meeting Summary
- Loss Reason Analysis
-
Examples include:
You can add up to 8 skills to an agent.
Note: Adding a skill does not bypass AI Studio Setup restrictions. Disabled tools, modules, fields, models, or knowledge sources cannot be enabled through a skill.
-
Add Tools: Tools determine what the Background Agent can use to access data, perform CRM actions, run custom functions, search external sources, or work with connected apps. Click **Add Tool** to add the required tools.
-
Tools can include:
- Workspace Tools
- Connected App Tools
- Custom Tools
-
Tools can include:
For each selected tool, configure its policy to either Allow the agent to use the tool or Require Approval before the action is executed.
-
Add CRM Data: CRM Data Access determines which live CRM data the Background Agent can read. Click Add CRM Data and select the required CRM data. The agent can only access the CRM data configured for it, subject to AI Studio Setup and the run user's CRM permissions. You can add data from modules such as:
- Contacts
- Companies
- Deals
- Activities
- Tickets
- Custom modules
To learn more about configuring CRM Data, see Configure CRM Data.
-
Configure User Inputs: User Inputs are available for Background Agents. Use them when the agent needs values to be provided before it runs.
-
Examples include:
- Selected records
- Date range
- Target owner
- Pipeline
- Review note
- Uploaded file
- Recipient user/team
- Threshold value
-
Examples include:
Supported input types may include:
- CRM record selection
- Text
- Number
- Date
- Boolean
Required inputs must be completed before a manual run. For Smart Flow automation, required inputs must be mapped before the automation can be activated.
-
Add Knowledge Sources: Knowledge Sources provide reference material for the Background Agent. They can be used for approved materials such as playbooks, policies, rules, and product documentation. Click Add Knowledge Source and add the required source.
-
Examples include:
- Sales playbook
- Qualification rules
- Renewal playbook
- Email guidelines
- Product comparison guide
- Support troubleshooting guide
-
Examples include:
-
Automate With Smart Flow: Background Agents can be automated using Smart Flow. You can configure the agent to run based on a schedule, trigger, or condition.For setup details, see How to Create Smartflow for Background Agent?
-
Examples include:
- Every weekday at 9:00 AM, check stale deals.
- When a lead is created, qualify it.
- When a high-value deal has no activity, create a follow-up task.
- Every Friday, summarize customer call insights from the week.
-
Examples include:
Smart Flow automation is optional. Background Agents can also be published without Smart Flow automation.
-
Configure Run Owner, Approvals, and Final Output: Configure the Run Owner to determine whose CRM permissions are used when the Background Agent runs.
You can use a dynamic run owner or select a specific run owner. Approval routing and final output routing determine who receives approval requests and final results. These settings do not change whose CRM permissions are used for the run.
- Configure User Access: Configure who can access the Background Agent after it is published. User Access determines who can use the published agent. Profile permissions continue to control other actions such as editing, publishing, or deleting the agent.
- Configure Model Settings: Configure the available model settings for the Background Agent as required. The workspace default model is used unless a specific model override is required and allowed by AI Studio Setup.To learn more about AI model configuration, see Configure AI Models.
- Before publishing, click Test to run the Background Agent in preview mode.
- Click Save to save the Background Agent as a draft. You can continue editing the draft before publishing it.
- Once the Background Agent is configured and tested, click Publish.
AI Studio validates the agent before publishing. The validation checks the required configuration, including fields, skills, tools, CRM data, knowledge, action policies, approvals, User Access, permissions, model settings, and other required setup restrictions.
- Once the validation is successful, the agent is published and becomes available based on the configured User Access.
Comments
0 comments
Article is closed for comments.