siteaiagent Documentation

Use siteaiagent to create an AI support agent for your website. The agent answers visitor questions from your website content and uploaded documents, captures qualified leads, and hands off conversations that need human review.

What siteaiagent does

siteaiagent is built for teams that receive repeated questions about pricing, features, onboarding, support, or service scope. It gives visitors fast answers, keeps the conversation on your site, and routes high-intent conversations to your team.

Answer visitor questions from approved content.
Capture contact details and buying intent inside the chat.
Hand off complex conversations with context.
Deploy a lightweight widget with a single script tag.

Core concepts

Agent

An agent is one independent support configuration for one website or product. It has its own knowledge sources, answer rules, lead form, widget appearance, and handoff settings.

Knowledge sources

Knowledge sources are the indexed content that the agent searches before answering. You can add content by crawling public pages or by uploading .txt and .md documents.

The widget is the floating launcher and iframe chat window that runs on your website. widget.js loads the launcher, passes configuration to the iframe, and keeps the chat UI isolated from your page.

Conversation

A conversation is one visitor session with the agent. It includes visitor questions, AI replies, collected lead details, and handoff status.

Documentation sections

Start with Quick Start to create and publish the first agent.
Use Agents to understand the workspace structure and agent settings.
Use Knowledge Sources before importing public pages or documents.
Use Answer Rules to shape tone, boundaries, and escalation behavior.
Use Widget when installing the hybrid script and iframe runtime.
Use Leads & Handoff to capture qualified visitors and route human follow-up.
Use Deployment for the final launch checklist.