Skip to main content
ChattyBox Logo

Welcome to ChattyBox

Turn your documentation into a source-cited AI chatbot that answers from your content, embeds with one script tag, and exposes the questions your docs still need to answer.

Fast path​

If you already have a docs site or help center, start with Getting Started, pick the right Installation Guide, and use Analytics to find unanswered questions.

  1. Create a ChattyBox project. Add your site URL or sitemap so the crawler can discover documentation pages.
  2. Test grounded answers. Ask real support questions and verify the cited source pages.
  3. Install the widget. Paste the script tag with data-api-key and data-api-url before </body> or use your framework's script loader.

Get running

Create a project, scrape your content, test responses, and launch the widget in minutes.

Install the widget

Embed ChattyBox on any site and customize colors, copy, placement, and language behavior.

Use an AI builder

Copy prompts for Cursor, Claude, v0, Bolt, Lovable, Replit, and similar tools.

Improve your docs

Track conversations, identify content gaps, and decide what documentation to write next.

What ChattyBox Does​

ChattyBox is an AI-powered chatbot platform that:

  • Scrapes your website and indexes documentation, guides, help centers, or knowledge bases.
  • Answers from retrieved content instead of generic model memory.
  • Shows source citations so visitors can verify answers and continue reading.
  • Refuses missing information instead of inventing unsupported answers.

Quick Start​

Get your chatbot running in under 5 minutes.

  1. Create an account and make your first project.
  2. Add your website URL so ChattyBox can crawl and vectorize your content.
  3. Install the widget using the path for your site platform.
  4. Run the launch checklist before publishing broadly.
  5. Review analytics to see what visitors ask and what content is missing.

Key Features​

Source-Grounded Answers​

Unlike generic chatbots, ChattyBox is designed to answer only from your scraped content. If the information isn't in your documentation, the bot can say it doesn't know. Answers include source citations for verification.

Source Citations​

Every answer includes clickable links to the source pages in your documentation. This builds trust and lets users verify information.

Analytics Dashboard​

Track what questions visitors are asking, identify content gaps, and improve your documentation based on real user needs.

Easy Customization​

Match the chat widget to your brand with custom colors, icons, and welcome messages, all from the dashboard with no code required.

How It Works​

  1. Scraping - Our crawler visits your website, extracts text content, and handles sitemaps automatically
  2. Vector Embeddings - Content is converted to high-dimensional vectors for semantic search
  3. RAG Response - When a user asks a question, we retrieve relevant chunks and generate an answer using only that context

1. Getting Started

Create the project and understand the basic launch flow.

2. Content Scraping

Control what gets indexed and keep multilingual content organized.

3. Widget Setup

Install, customize, and localize the visitor-facing chat experience.

4. Troubleshooting

Fix missing widget, API key, origin, language, citation, and stale content issues.

Pricing​

ChattyBox offers a free tier with generous limits for testing, plus affordable paid plans as you grow. See our pricing page for details.

Frequently Asked Questions​

What content can ChattyBox index?​

ChattyBox can index website pages, documentation, help centers, CMS content, knowledge bases, and sitemap-discovered pages. The crawler extracts readable text and headings so the assistant can answer from source content instead of generic model memory.

How do source citations work?​

When a visitor asks a question, ChattyBox retrieves relevant passages from indexed pages and can include links to the source pages used for the response. Citations help visitors verify answers and continue reading the official documentation.

What should I test before launch?​

Ask real support and documentation questions, confirm the cited pages are correct, check fallback behavior when information is missing, and run through the launch checklist before publishing the widget broadly.

Need Help?​