Bots

Bots are the core foundation of ChatBotKit's conversational AI platform. They represent tangible, customized conversational AI solutions designed and deployed by users. Each bot is a unique instance that combines artificial intelligence models with user-defined behavior, knowledge bases, and capabilities to create meaningful, contextual interactions with end users.

Overview

A ChatBotKit bot is more than just a language model - it's a comprehensive conversational system built from multiple interconnected components:

• Backstory: Natural language instructions that define the bot's personality, expertise, and behavior patterns
• Model: The underlying AI model (GPT, Claude, etc.) that generates responses
• Datasets: Curated knowledge bases that provide contextual information for accurate, relevant responses
• Skillsets: Custom abilities and functions that extend the bot's capabilities beyond standard conversation
• Configuration: Advanced settings for privacy, moderation, and behavioral control

This modular architecture allows you to create bots tailored to virtually any use case - from customer service and technical support to educational assistance and more.

Creating a Bot

Step-by-Step Guide

The bot creation process in ChatBotKit is designed to be intuitive while providing powerful customization options. Here's how to create your first bot:

Initiate Bot Creation

Navigate to the ChatBotKit dashboard and click the Create Bot button. This launches the bot configuration wizard, which guides you through each configuration stage.

Provide Basic Information

Fill in the following details:

• Name: A descriptive identifier for your bot (e.g., "Customer Support Assistant", "Product Knowledge Bot")
• Description: A brief explanation of your bot's purpose and primary function (optional but recommended for organization and agent-to-agent communication)

These details help you organize and identify your bots within the dashboard, especially when managing multiple bots across different projects.

Define the Backstory

The backstory is the most critical configuration element. It serves as natural language programming instructions that shape every aspect of your bot's behavior.

A well-crafted backstory should include:

• Role and Purpose: Clearly state what the bot does and its primary responsibilities
• Personality and Tone: Define communication style (formal, conversational, technical, friendly, etc.)
• Expertise and Knowledge Scope: Specify areas of expertise and boundaries of what the bot claims to know
• Behavioral Guidelines: Include instructions on how to handle edge cases, escalations, and limitations
• Response Format: Specify preferred response structures, lengths, and styles

Example backstory for a customer support bot:

Associate Datasets (Optional)

Datasets provide your bot with curated knowledge bases to enhance response accuracy and relevance. By connecting datasets, your bot can:

• Reference specific product information, documentation, or FAQs
• Provide data-driven answers grounded in your organization's knowledge
• Reduce hallucinations by grounding responses in verified content
• Maintain consistency across conversations

During conversations, ChatBotKit uses retrieval-augmented generation (RAG) to search these datasets and incorporate relevant information into bot responses.

When to use datasets:

• Your bot needs to reference specific company information (product specs, policies, procedures)
• You want to minimize hallucinations with verified data
• You need to provide consistent, accurate information across many conversations
• Your knowledge base changes frequently and requires centralized updates

Associate Skillsets (Optional)

Skillsets extend your bot's capabilities beyond conversation. They enable your bot to perform specific tasks, integrate with external tools, or access specialized functions.

Common skillset use cases include:

• Image Generation: Enable the bot to create images based on user requests
• Web Fetching: Allow the bot to retrieve and summarize current web content
• Data Processing: Perform calculations, data transformations, or analysis
• API Integration: Connect to external services (payment processors, booking systems, etc.)
• Code Execution: Execute code snippets or scripts in response to user requests

Each skillset contains multiple abilities, to create rich, multi-functional conversational agents.

Model selection:

Choose the underlying AI model that powers your bot. ChatBotKit supports models from multiple providers:

• OpenAI: GPT-5 (latest), GPT-4.1, GPT-4o, and cost-efficient variants (mini, nano)
• Anthropic: Claude 4.5 Sonnet, Claude 4 Opus, and other Claude models
• Other Providers: Mistral, Groq, DeepSeek, and others

Model selection should balance:

• Capability Requirements: Complex reasoning tasks benefit from larger models
• Cost Considerations: Smaller models (mini, nano variants) reduce API costs
• Speed Requirements: Lightweight models offer faster response times
• Specialized Features: Some models excel at coding, math, or multimodal tasks

Privacy settings:

Enable privacy protection to automatically anonymize personally identifiable information (PII) in conversations:

• PII Detection and Anonymization: Detects names, emails, phone numbers, addresses, social security numbers, and other sensitive data
• Entity Handling: Converts detected PII into anonymized tokens while maintaining conversation coherence
• Transparent Processing: You receive information about detected entities and anonymization transformations

When enabled, privacy features ensure compliance with regulations like GDPR and CCPA while protecting user data throughout the bot's lifecycle.

Moderation settings:

Content moderation is essential for maintaining safe, appropriate interactions:

• Inbound Scanning: All incoming user messages are scanned before processing
• Outbound Scanning: All bot-generated responses are scanned before delivery
• Language Detection: The system recognizes harmful, hateful, and inappropriate language
• Automatic Refusal: Flagged content is automatically blocked with a default refusal message

When content is flagged, you receive email notifications and can review flagged messages in the conversations dashboard.

Bot Configuration Summary

Deployment and Integration

Bots created in ChatBotKit can be deployed across multiple channels:

Supported Integration Channels

• Widget: Embed directly on websites
• Slack: Integrate with Slack workspaces
• Discord: Deploy bots for Discord communities
• WhatsApp: Connect to WhatsApp Business API
• Messenger: Integrate with Facebook Messenger
• Telegram: Deploy as Telegram bots
• Twilio: Connect to telephony and SMS systems
• Email: Interact via incoming emails
• Custom APIs: Build custom integrations via REST API

Each channel may have specific configuration requirements and feature support. Refer to the integration-specific documentation for detailed setup instructions.

Troubleshooting