ZChat ZChat 2.0 Self-Hosted Live Chat
AI + human

AI chatbot with human handoff

The best support experience isn't AI or humans - it's both, in the right order. ZChat's chatbot answers common questions from your knowledge base, and the moment a conversation needs a person, it hands off with the full transcript attached. No dead ends, no "please repeat that."

One thing to know up front: the chatbot ships as an API. The widget you drop on your site connects visitors to your human agents out of the box - putting the AI in front of them is an integration step you control. Details below.

Try It Free About the AI Chatbot When Not to Use AI

What's turnkey, and what stays your choice

We'd rather you know exactly how this ships than have you find out after downloading:

Working the moment you install it: the visitor widget talking to your human agents, offline message capture, the knowledge base, and AI reply suggestions and summaries and tagging for your team.

Two settings away: the visitor-facing assistant. The chatbot and its widget surface are separate switches, both off by default - deliberately, so enabling AI for your agents never silently puts a bot in front of customers. Turn both on and the widget offers the assistant whenever no agent is available, with a per-visitor answer budget, per-IP rate limiting, and a handoff to your team on every failure path. "Talk to a human" always works, from the first message.

An API when you want more: the same conversation engine is exposed over HTTP, so you can put the AI in front of the live queue, inside your mobile app, or anywhere else the stock placement doesn't cover.

1. The AI answers first

A visitor asks a question and the chatbot responds immediately, grounded in your own knowledge base articles rather than making things up. Routine questions - hours, pricing, how-tos - get an instant, accurate answer at any time of day.

2. It recognizes when to step aside

When the visitor asks for a person, uses a handoff keyword you've configured (like refund, cancel, or human), or asks something the knowledge base can't answer confidently, the assistant stops trying to force an answer and routes the conversation to a live agent.

3. The agent takes over with full context

The conversation appears in the operator console with the entire prior exchange - the visitor's questions and the AI's answers - already attached. The agent reads in, replies, and the customer never has to repeat what they already said to the bot.

4. You stay in control of the AI

You set the system prompt, the handoff keywords, and which model runs - a local Ollama model so nothing leaves your network, or your own OpenAI/Anthropic keys. The whole flow runs on your self-hosted server against your own data.

Why handoff matters

Deflect the routine, escalate the rest

A bot that can't hand off frustrates people; agents who answer every "what are your hours" burn out. Handoff gives you both: the AI absorbs the repetitive volume, and your team spends its time on the conversations that actually need a human - with the context to resolve them fast.

The AI handles

Repeated questions answered from your knowledge base, 24/7, in the visitor's language, without adding headcount.

The human handles

Judgment calls, upset customers, edge cases, and anything requiring account access - handed the full transcript so there's no cold start.

Does the chatbot work out of the box?
Not with the stock widget. The chatbot ships as an API - the engine, the knowledge-base grounding and the handoff logic are all included and documented, but the widget you drop on your site connects visitors to your human agents. Putting the AI in front of visitors is an integration step. Reply suggestions, summaries and tagging, by contrast, work as soon as you enable them.
What is AI chatbot human handoff?
Human handoff is when an AI chatbot recognizes it can't or shouldn't answer and passes the conversation to a live agent, along with the full transcript so the customer never has to repeat themselves. It combines AI answers for common questions with a human for anything that needs judgment.
How does the chatbot decide when to hand off to a human?
ZChat hands off when the visitor asks for a person, when their message contains configured handoff keywords (like refund, cancel, or human), or when the assistant cannot answer confidently from the knowledge base. You control the keywords and the system prompt.
Does the agent see the AI conversation?
Yes. When the chat is handed off, the agent picks it up in the console with the entire prior conversation - including the AI's answers - already attached, so they have full context immediately.
Can the AI run privately without sending data out?
Yes. ZChat's chatbot can run on a local Ollama model so the AI runs on your own server and nothing leaves your network, or you can connect your own OpenAI or Anthropic API keys. It answers grounded in your own knowledge base either way.
Self-hosted AI

AI first response, human when it counts

The chatbot and the handoff are built in - no per-resolution AI fees from ZChat. Run it on your own server with a local model, grounded in your knowledge base, with your team one step away.

Answers from your knowledge base

Keyword and low-confidence handoff

Full transcript passed to the agent

Local Ollama or your own provider keys

Set Up the AI Chatbot Download Free Evaluation
Owned customer support software

Deploy live chat on your own terms, not on someone else's pricing model.

ZChat gives you the installable server, web dashboard, website widget, and desktop agent tools in one self-hosted product you buy once and keep. Run it on infrastructure you trust and connect AI only if and how you want it.

Download ZChat See Product Demo

Deployment

Install on Windows or Linux, behind IIS or Nginx, in a VM, or in Docker if that fits your stack.

Commercial model

One-time purchase, perpetual license, and no monthly per-agent bill attached to growth.

AI Flexibility

Use Ollama locally or connect OpenAI and Anthropic with your own provider accounts.