Skip to main content

Using Team Insights

Introduction​

Team insights in KaibanJS allow you to provide your agents with a shared knowledge base of historical data, patterns, and experiences. This feature enables agents to make more informed decisions and provide context-aware responses based on past experiences and established patterns.

Using AI Development Tools?

Our documentation is available in an LLM-friendly format at docs.kaibanjs.com/llms-full.txt. Feed this URL directly into your AI IDE or coding assistant for enhanced development support!

How to Implement Team Insights​

To use team insights, you need to provide an insights string when creating your team. This string can contain any relevant information that you want your agents to consider during task execution.

Basic Usage​

Here's a simple example of how to create a team with insights:

import { Agent, Task, Team } from 'kaibanjs';

const team = new Team({
  name: 'Customer Service Team',
  agents: [/* your agents */],
  tasks: [/* your tasks */],
  insights: `
Customer Service History (2023):
1. Most common inquiries relate to account access (45% of tickets)
2. Peak support hours: 9 AM - 11 AM EST
3. Average response time: 2.5 minutes
4. Customer satisfaction highest with video tutorials
5. Common pain point: password reset process
  `
});

Structured Insights​

For more complex scenarios, you can structure your insights to make them more organized and easier for agents to reference:

const team = new Team({
  name: 'Travel Experience Team',
  agents: [profileAnalyst, recommendationAgent],
  tasks: [analysisTask, recommendationTask],
  insights: `
Passenger Profile: Michael Zhang (FF-789321)
Travel History and Preferences (2023):
1. Seating: Window seats (87% of flights), Row 11 preferred
2. Dining: Asian Vegetarian meals (9/10 long-haul flights)
3. Airport Services: Terminal 4 lounge user (avg. 2.5hr stays)
4. Entertainment: Dissatisfied with 787 fleet options
5. Past Issues: Delays on DXB-892, requested better updates

Flight Statistics:
- Total Flights: 24
- Long-haul: 10
- Domestic: 14
- On-time Rate: 92%

Customer Feedback:
- Survey ID: CS-982
- Satisfaction Score: 4.2/5
- Key Comments: "Great service, needs better delay updates"
  `
});

Best Practices​

  1. Structure Your Insights

    • Organize information into clear categories
    • Use bullet points or numbered lists for better readability
    • Include specific identifiers and metrics when available

  2. Keep Information Relevant

    • Include only information that agents need for their tasks
    • Update insights regularly to maintain relevance
    • Remove outdated or irrelevant information

  3. Include Specific Details

    • Use concrete numbers and statistics
    • Reference specific IDs or identifiers
    • Include dates and timeframes

  4. Format for Readability

    • Use clear section headers
    • Break down long paragraphs
    • Highlight key information

Common Use Cases​

  1. Customer Service

    • Customer interaction history
    • Common issues and solutions
    • Peak service times
    • Customer satisfaction metrics

  2. Travel Services

    • Passenger preferences
    • Travel history
    • Service ratings
    • Common complaints

  3. Project Management

    • Team performance metrics
    • Project success rates
    • Common bottlenecks
    • Best practices

  4. Product Development

    • User feedback history
    • Feature usage statistics
    • Common feature requests
    • Bug patterns

Example: Full Implementation​

Here's a complete example showing how insights can be used in a customer service context:

import { Agent, Task, Team } from 'kaibanjs';

// Define agents
const analysisAgent = new Agent({
  name: 'Alice',
  role: 'Customer Profile Analyst',
  goal: 'Analyze customer history and identify key patterns',
  background: 'Customer Behavior Analysis',
});

const supportAgent = new Agent({
  name: 'Bob',
  role: 'Support Specialist',
  goal: 'Provide personalized support solutions',
  background: 'Customer Service and Problem Resolution',
});

// Define tasks
const analysisTask = new Task({
  description: 'Analyze customer profile and recent interactions',
  expectedOutput: 'Detailed analysis of customer needs and patterns',
  agent: analysisAgent,
});

const supportTask = new Task({
  description: 'Create personalized support plan based on analysis',
  expectedOutput: 'Support plan with specific recommendations',
  agent: supportAgent,
  isDeliverable: true,
});

// Create team with insights
const team = new Team({
  name: 'Customer Support Team',
  agents: [analysisAgent, supportAgent],
  tasks: [analysisTask, supportTask],
  insights: `
Customer: John Smith (ID: CS-45678)
Support History (Last 90 Days):
1. Ticket #T-892: Login issues with mobile app
   - Resolution: Guided through app reinstallation
   - Satisfaction: 4/5

2. Ticket #T-923: Billing inquiry
   - Resolution: Explained premium features
   - Satisfaction: 5/5

Usage Patterns:
- Primary platform: Mobile (iOS)
- Feature usage: Heavy user of analytics dashboard
- Login frequency: Daily
- Subscription: Premium tier

Previous Feedback:
- Appreciates quick responses
- Prefers video tutorials over text
- Has mentioned interest in API access
  `,
  env: { OPENAI_API_KEY: process.env.OPENAI_API_KEY }
});

// Start the workflow
team.start()
  .then(output => {
    console.log('Support plan:', output.result);
  })
  .catch(error => {
    console.error('Error:', error);
  });

Conclusion​

Team insights are a powerful feature in KaibanJS that enable you to create more context-aware and effective multi-agent systems. By providing your agents with relevant historical data and patterns, you can significantly improve the quality and relevance of their outputs.

We Love Feedback!

Is there something unclear or quirky in the docs? Maybe you have a suggestion or spotted an issue? Help us refine and enhance our documentation by submitting an issue on GitHub. We're all ears!