Introduction

What is branchly?

branchly is the AI communication layer for your website and digital touchpoints. Built on top of a single centralized knowledge base, branchly combines semantic-hybrid instant search, guided navigation, smart AI forms, and generative AI chat (Agentic RAG) in one unified no-code platform.

Instead of disconnected point solutions, every interface in the branchly dashboard shares the same knowledge base, AI Actions, and analytics engine:

Chat & Chat Widget: Conversational AI agent delivering source-cited answers, page-aware suggested questions, and active process automation via AI Actions.
Search (AI Search): Semantic-hybrid search (combining dense vector search, sparse BM25 keyword search, and reranking) with AI-generated direct answers and keyword highlighting.
Navigator: An interactive self-service interface blending search and node-based topic navigation to guide visitors to the right solution.
Form (AI Form): A smart form interface that answers routine questions directly from your knowledge base before submission and routes unresolved requests to the right form.
Headless API, Voice & WebMCP: Open interfaces to integrate branchly into custom apps, messaging channels (such as WhatsApp or Slack), voice/telephony agents, and browser-native AI agents.

How branchly Works

branchly operates as a continuous closed loop — from ingesting your existing content to answering visitor inquiries in real time and turning interactions into actionable insights:

1. Connect & Structure Your Knowledge

Using Data Sources, you automatically ingest existing content into your Knowledge Base — via the Website Crawler, File Uploads (PDF, CSV), WordPress, HelpSpace, OpenAPI specifications, or Webhooks. You can also create and edit individual nodes manually in the WYSIWYG Node Editor, manage multiple locales, and prioritize key content via score_boost.

Because branchly is built on a modern Agentic RAG architecture (Retrieval-Augmented Generation) rather than static model fine-tuning, new or updated content is indexed and available across all interfaces immediately in real time.

2. Configure AI Agents, Prompts & AI Actions

For each User Interface, you configure the agent workflow across transparent processing stages (Pre-Processing, Retrieval, Routing, Tools, Answering, and Post-Processing):

Source-Grounded Answers: The AI answers strictly from your approved knowledge base and active tools, automatically citing the source URLs used.
AI Actions (Tools): With AI Actions, you extend the AI agent beyond information retrieval — enabling it to call external REST APIs, custom or pre-built MCP Servers, interactive forms, Calendly scheduling, Google Maps directions, or domain integrations in real time.
Programmatic Management via MCP & Skills: Through the branchly MCP Server and official branchly Skills, you can also inspect, configure, audit, and optimize your application directly from AI assistants such as ChatGPT, Claude, Langdock, Claude Code, or OpenCode.

3. Embed & Customize User Interfaces

All User Interfaces are built as lightweight Web Components. You can embed them into your website in minutes using a simple HTML/JavaScript snippet (see How to Embed a User Interface) and tailor colors, typography, greetings, and behavior in the dashboard or via custom CSS styling.

With built-in multilingual support and optional browser-language detection (see Application Settings), a single setup automatically serves visitors in 130+ languages.

4. Analyze Interactions & Continuously Optimize

Every visitor interaction is tracked completely cookie-free and analyzed in real time:

Insights: Visualizes active sessions by interaction and embed type, top search queries, clicked URLs, sentiment distribution, and automated topic/intent classifications — complemented by scheduled weekly email reports.
Conversations: Provides full session-level transparency — including complete message transcripts, executed AI Actions, answer resolution classifications (Complete, No Knowledge, Out of Scope), and exact cited knowledge base nodes.

This gives your team immediate visibility into what visitors actually ask, where knowledge gaps (No Knowledge) exist, and which content to expand next.

Explore the Documentation

Main Concepts: Overview of Organizations, Applications, Knowledge Base nodes, Data Sources, and AI Actions.
Manage your application: Deep dives into User Interfaces, Knowledge Base, Data Sources, AI Actions, Insights, Conversations, Application Settings, and Webhooks.
MCP & Skills: Connect the MCP Server and use branchly Skills.
Guides: Practical guides on Embedding Interfaces, Custom CSS Styling, and Prompting.