Agents Overview

Introduction to Agents

ChatBotKit SDK provides a powerful and flexible way to create, manage, and deploy intelligent conversational agents. This guide will help you understand the core concepts and show you how to effectively use the SDK to build sophisticated chatbots and AI assistants.

Overview

The ChatBotKit SDK enables you to create conversational agents by combining several key components:

  • Bots: The core entity that represents your conversational agent
  • Skillsets: Collections of abilities that define what your bot can do
  • Datasets: Knowledge bases that your bot can reference
  • Conversations: Interactive sessions between users and your bot

Core Components

Bots

Bots are the foundation of your conversational agents. Each bot can be configured with:

  • A specific language model (e.g., 'gpt-4o')
  • A backstory that defines its personality and behavior
  • Associated skillsets and datasets
  • Custom metadata for organization and filtering
const bot = await cbk.bot.create({
  name: 'Customer Service Bot',
  description: 'AI assistant for customer support',
  model: 'gpt-4o',
  backstory: 'You are a helpful customer service representative.',
  skillsetId: 'customer-support-skills',
  datasetId: 'product-knowledge-base'
})


javascript

Skillsets and Abilities

Skillsets allow you to define specific capabilities for your bot. Each skillset can contain multiple abilities that represent different functions or tasks your bot can perform:

// Create a skillset for customer support
const skillset = await cbk.skillset.create({
  name: 'Customer Support Skills',
  description: 'Essential abilities for customer service'
})

// Add specific abilities to the skillset
await cbk.skillset.ability.create(skillset.id, {
  name: 'Process Refund',
  description: 'Handle customer refund requests',
  instructions: 'Follow company refund policy guidelines...'
})


javascript

Datasets

Datasets serve as knowledge bases for your bots, allowing them to access and reference specific information:

const dataset = await cbk.dataset.create({
  name: 'Product Knowledge Base',
  description: 'Product information and FAQs',
  searchMaxRecords: 5,
  searchMinScore: 0.7
})

// Add records to the dataset
await cbk.dataset.record.create(dataset.id, {
  text: 'Product documentation and specifications...'
})


javascript

Conversations

Conversations represent interactive sessions between users and your bot. They can be configured with specific models, skillsets, and datasets:

const conversation = await cbk.conversation.create({
  model: 'gpt-4o',
  botId: 'your-bot-id',
  skillsetId: 'customer-support-skills',
  datasetId: 'product-knowledge-base'
})


javascript

Best Practices

  1. Modular Design: Break down your bot's capabilities into discrete skillsets and abilities for better organization and reusability.
  2. Knowledge Management: Use datasets effectively to manage and update your bot's knowledge base without changing the core bot configuration.
  3. Conversation Management: Implement proper conversation handling with appropriate backstories and context to maintain consistent bot behavior.
  4. Performance Monitoring: Utilize the upvote/downvote system to track bot performance and user satisfaction.

Getting Started

To begin using the ChatBotKit SDK, first authenticate with your API key:

import { ChatBotKit } from '@chatbotkit/sdk'

const cbk = new ChatBotKit({
  secret: process.env.CHATBOTKIT_API_KEY
})


javascript

From there, you can create and manage your conversational agents using the components described above.

Additional Resources

Go to cbk.ai