AI agents are intelligent assistants pre-configured to help you with your sales process. You can configure every aspect of them to fit your needs. This guide covers the core concepts of agents in Bigmind.
What are AI Agents?
AI Agents in Bigmind are specialized assistants that help sales teams by providing intelligent recommendations, automating tasks, and maintaining context across your entire sales process. Each agent is configured for a specific purpose - whether it's managing accounts, working deals, or coaching sales reps.
Agent Configuration
When configuring an agent, you have access to several configuration tabs that define how your agent behaves:
Agent Settings
The Agent Settings tab contains the core configuration for your AI agent:
Agent Name
The display name for your agent that appears throughout the interface. Choose a descriptive name that reflects the agent's purpose, such as "Account AI agent" or "Deal AI agent".
Description
A brief description of what this agent does. This helps team members understand the agent's role and capabilities. For example: "Helps CSMs, AMs, AEs manage account health, growth potential and unlock strategic opportunities".
System Prompt
The system prompt is the most critical configuration - it defines your agent's behavior, personality, and capabilities. This is where you specify:
- Role and Introduction: Define what the agent is and its primary purpose
- Core Responsibilities: List the specific tasks the agent should handle
- Communication Style: Set the tone and approach for interactions
- Guidelines: Establish rules and best practices for the agent to follow
Example system prompt structure:
<introduction>
You are an Account AI Agent, a specialized sales assistant focused on account-level strategy and relationship management.
</introduction><core_responsibilities>
- Analyze account health, growth potential, and strategic opportunities
- Provide insights on account expansion, renewal, and upselling strategies
- Identify key stakeholders and decision-makers within the account
</core_responsibilities><communication_style>
- Be strategic and business-focused in your recommendations
- Provide data-driven insights with clear rationale
- Use professional language appropriate for C-level discussions
</communication_style>
Knowledge
The Knowledge tab allows you to provide your agent with context-specific information from your Library to enhance its responses. You can add Products, ICPs, Case Studies, Playbooks, and Documents.
When you add knowledge items to an agent, the agent can reference this information during conversations to provide more accurate and relevant responses. For example, adding product information allows the agent to recommend specific products, while adding playbooks helps the agent follow your established sales methodology.
Learn more about how to organize your knowledge in the Library documentation.
Tools
Tools extend your agent's capabilities by allowing it to perform specific actions and access data. Tools can interact with your CRM, send emails, research companies, and much more.
Learn more about how to configure tools for your agents.
MCPs (Model Context Protocol)
MCPs provide advanced capabilities and integrations for your agents. They extend your agent's functionality beyond basic tools and can maintain state across multiple interactions.
Learn more about how to use MCPs with your agents.
Background vs Foreground Agents
Bigmind supports two modes of agent operation, and you can configure both for the same agent using the toggle switch in the agent configuration:
Foreground Agents (Interactive)
Foreground agents are the agents users directly interact with in chat conversations. When a user opens a chat with an Account, Deal, Contact, or Lead, the foreground agent for that object type becomes available to assist.
Key characteristics:
- User-initiated: Respond to user questions and requests in real-time
- Conversational: Provide recommendations, insights, and guidance through natural dialogue
- Context-aware: Have access to the specific CRM object the user is viewing
- Interactive: Can use tools to fetch data, draft emails, and perform actions on behalf of the user
Common use cases:
- Help sales reps prepare for meetings by summarizing account history
- Answer questions about deal status and next steps
- Draft follow-up emails based on meeting notes
- Suggest relevant case studies or products for a prospect
- Provide coaching and guidance during sales conversations
Background Agents (Automated)
Background agents work automatically behind the scenes, monitoring events and processing data without user interaction. They're triggered by specific events in your workflow, such as when a meeting ends or when a deal moves to a new stage.
Key characteristics:
- Event-driven: Automatically triggered by system events (meeting completions, CRM updates, etc.)
- Autonomous: Work without user interaction to analyze data and suggest actions
- Batch processing: Can handle multiple events efficiently using workflows
- Separate configuration: Have their own system prompt, tools, and MCPs optimized for automation
Common use cases:
- Analyze meeting transcripts and suggest CRM field updates
- Monitor deal progress and flag risks or opportunities
- Track warnings and update deal summaries automatically
- Identify action items from conversations and create follow-up tasks
- Update framework assessments based on new information
Background Agent Configuration
Background agents have their own separate configuration that's optimized for automated processing:
- Internal System Prompt: Focused on analysis and data extraction rather than conversation
- Internal Tools: Specialized tools for CRM updates, framework assessments, and data processing
- Internal MCPs: Advanced capabilities for complex data operations
- Internal Knowledge: Context-specific information for automated decision-making
Example Background Agent Workflow
When a meeting ends, a background agent might:
- Receive the trigger: The meeting completion event triggers the background agent workflow
- Analyze the transcript: Process the meeting transcript to extract key information
- Identify updates: Determine which CRM fields should be updated based on the conversation
- Suggest changes: Generate structured suggestions for field updates
- Flag insights: Identify any risks, opportunities, or action items discussed
- Update summaries: Refresh deal summaries and framework assessments with new information
This entire process happens automatically in the background, ensuring your CRM stays up-to-date without manual data entry.
Testing Your Agent
The right side of the agent configuration interface provides a testing environment where you can validate your agent's behavior before deploying it to your team.
Test Environment Features
- Object Selection: Select a specific CRM object (Account, Contact, Deal, Lead) to test against
- Real Data: Test with actual data from your CRM to see how the agent performs in realistic scenarios
- Tool Execution: Verify that tools are working correctly and returning expected results
- Knowledge Access: Confirm that the agent can access and use the knowledge you've configured
- Prompt Refinement: Iterate on your system prompt based on the agent's responses
Testing Best Practices
- Test with diverse objects: Try different types of accounts, deals, or contacts to ensure consistent behavior
- Verify tool calls: Check that the agent is calling the right tools at the right times
- Review responses: Ensure the agent's tone and communication style match your expectations
- Check knowledge usage: Confirm the agent references the correct products, case studies, or playbooks
- Iterate and refine: Use test results to improve your system prompt and configuration
This testing capability ensures your agent works as expected before deploying it to your team, reducing the need for adjustments after rollout.
Agent Types by Object
Bigmind agents are typically configured for specific CRM object types:
- Account Agents: Focus on account-level strategy, relationship management, and expansion opportunities
- Deal Agents: Help manage deal progression, identify risks, and suggest next steps to close
- Contact Agents: Assist with contact-level interactions, relationship building, and communication
- Lead Agents: Support lead qualification, nurturing, and conversion processes
Each agent type has access to object-specific tools and context, allowing them to provide highly relevant assistance for that particular stage of the sales process.