Skip to main content

Chatbots Overview

What It Is​

BotBat Chatbots are AI chatbots.
They are optimized for retrieval-based answers from your configured knowledge sources, not for full agent orchestration.

The chatbot module is focused on:

  • Knowledge collection (text, website, Q&A, files)
  • Knowledge-grounded answers
  • Language behavior controls
  • Safe lifecycle management

Agentic capabilities (conditional tools, human-in-the-loop, advanced orchestration) should be implemented in Workflows as a separate layer.

Scope Boundary​

Use Chatbots when you need:

  • Fast knowledge-driven responses
  • Controlled and predictable support/FAQ behavior
  • One bot that can answer from configured sources

Use Workflows when you need:

  • Multi-step tool orchestration
  • Conditional tool routing
  • Human approvals/escalations
  • Complex action pipelines

Knowledge Sources​

Each chatbot can use one or more of these sources:

  • Text content
  • Website URL
  • Q&A pairs
  • Uploaded/selected files

In the console, these appear as vertical source tabs with interactive completion indicators:

  • Filled source: green check icon
  • Not filled: orange circle icon

At least one source is required before create/update.

Validation Limits​

Current limits are enforced in frontend and backend:

  • knowledgeContent: up to 200,000 characters
  • knowledgeFileKeys: up to 50 files
  • listOfQA: up to 300 completed Q&A pairs

Language & Translation Mode​

  • language: default bot language (usually follows UI language at creation)
  • strictToLanguage:
    • false: translation mode enabled (bot replies in user message language)
    • true: language-restricted behavior

Lifecycle Status​

Knowledge base lifecycle statuses:

  • Draft
  • Active
  • Training
  • OnHold

Behavior note:

  • OnHold bots must not respond.

Rebuild Behavior​

Any chatbot create/update that changes knowledge fields triggers a knowledge-base rebuild in downstream services.
Status should be used operationally to reflect training and serving readiness.

Common Pitfalls​

  • Treating Chatbots as full agents (they are AI chatbot-first)
  • Creating bots with no real knowledge source
  • Overloading with uncurated files/Q&A
  • Ignoring OnHold as an operational stop state

Screens In This Flow​

List view showing chatbots with status badges
Create chatbot dialog with name and type selection
AI chatbot configuration panel with knowledge sections
Chatbot detail page with configuration tab open