Chattermate the open source ai chatbot

ChatterMate is a free, open-source AI customer support chatbot that claims it can be self-hosted or started for free directly, and can provide accurate, cited answers instantly based on your own documents.

Purpose It is designed for customer support scenarios, placing AI and human teams together in a shared inbox to work collaboratively, letting AI handle common questions and humans step in when needed, thereby reducing tickets and improving response efficiency.

Main capabilities

• Knowledge indexing and cited answers: scans, splits, and indexes your documents and pages, then answers based on that content with citations, without tagging or preparing training data.
• Unified inbox: every conversation goes into a shared inbox, with AI handling and human handling displayed side by side; supports open, pending, and closed views, cross-channel filtering, search, reassignment, and human takeover at any time.
• One-click handoff to a human with context retention: with one click, a conversation can be handed off to a human while preserving full context; Jira tickets can also be created directly from any conversation.
• Multi-capability agent: a single agent can handle chat topics, lead capture, human handoff, omnichannel, FAQ, help center, Jira, and Zendesk, etc.
• Editable prompts: agents can be named, tone can be set, and when to escalate or request feedback can be decided, all through editable prompts.
• Visual multi-step flows: visually build steps such as conditions, forms, messages, and human handoff, without engineers.
• MCP tool connections: after connecting MCP tools, the agent can query orders, initiate refunds, create tickets, and call your own APIs during conversations.
• Lead capture: while answering questions, the agent asks a few smart questions at the right time and captures leads; it helps first and asks later, routes to sales, Slack, or your CRM after filtering by intent, and groups every visitor, lead, and customer into the same deduplicated profile.
• Help center: automatically builds the same documents used to train the agent into a branded, searchable help center, keeps it in sync, and puts an AI answer at the top of every search, enabling customer self-service and reducing tickets.

How to use Official three-step process:

1. Create an account: sign up with email and go straight into the dashboard. The first 300 chats are free.
2. Feed it and shape it: upload a PDF or enter a URL, write the agent's persona prompt, and preview answers in real time.
3. Paste one line of code to go live: copy the widget code snippet to your website and go live in less than two minutes.

In addition, it can also be operated through the terminal: use a single CLI to manage signup, agents, knowledge, and workflows; every command supports --json, making it easy to script in CI; you can connect the MCP server to your AI client (using a personal access token) and run the entire system on your own infrastructure.

Use cases

• Automated customer support for a website or app, answering common questions based on your own documentation
• Support teams that need AI and human collaboration with unified inbox management
• Teams that want to capture and qualify sales leads through conversations and connect them to a CRM or Slack
• Organizations that want to self-host and control their own data and models
• Teams that need to turn support knowledge into a self-service help center at the same time

Widget and appearance Six preset styles are provided (such as Glass, terminal, playful, docs, ask-me-anything), and you can choose a brand color. The widget on the right updates in real time, showing exactly how it will look once embedded; colors, shapes, fonts, and position can all be customized.

Limitations and FAQ

• Official description: you can go live in less than two minutes; after signing up, it points you to the docs, you paste one line of code, and the agent automatically indexes your knowledge.
• When the AI cannot answer, it detects whether the customer is stuck or frustrated and hands the conversation with full context to your team, or handles it on its own according to your preference.
• The platform is fully open source, so you can inspect it, self-host it, and bring your own model, or use the cloud to avoid operations work.
• On pricing, the materials only mention that the free version covers the first 300 chats; the remaining prices and plans are not currently provided.
• The open source license is Apache-2.0, and the source code, license, and release history can be viewed on GitHub.
• The platform also provides self-verification information such as security and data controls and its approach to AI research, but the specific details are not currently provided.