Hexabot is the open source chatbot framework for teams that need full control over their conversational AI: a visual flow editor, NLU training pipeline, LLM integrations, and multi-channel deployment, all running on infrastructure you own with no per-conversation billing.
The Problem
Dialogflow charges per request and processes your conversation data through Google's cloud. Botpress has shifted toward a managed cloud business model. For teams with data residency requirements, industries where customer conversation data is sensitive, or organizations that want to avoid vendor lock-in on a customer-facing product, these platforms create ongoing compliance and cost risk.
How Hexabot Solves It
Hexabot ships as a self-hosted stack: Node.js API backend, React-based visual builder, and NLU service. Design conversation flows with the visual editor, train intent detection on your own utterances, plug in an LLM for generative fallback responses, and deploy to web chat, WhatsApp, Facebook Messenger, or custom channels, all from a single platform running on your servers.
Key Features
- Visual flow builder with drag-and-drop blocks for messages, conditions, variables, and API calls
- Built-in NLU: train intent classifiers on your own data without a third-party NLU service
- LLM plugin support for generative responses and RAG-based knowledge retrieval
- Multi-channel: web widget, WhatsApp, Facebook Messenger, and custom channel adapters
- Multilingual: train NLU models and serve flows in multiple languages simultaneously
- Role-based access control for team-managed bot administration
Who It's For
Hexabot is best for customer support teams deploying AI-assisted chatbots with full data control, organizations in regulated industries where conversation data cannot leave their own infrastructure, and developers building multilingual customer-facing bots without a per-message SaaS bill.
Compared to Dialogflow
Unlike Dialogflow or Botpress cloud, Hexabot is fully self-hosted: conversation data stays on your servers, there is no per-request billing to Google's infrastructure, and you train NLU models on your own utterances without a third-party NLU service.
