AnswerDesk AI Product Knowledge
Inspect the original source, the normalized retrieval document, and the chunks stored for vector search.
AnswerDesk AI Product Knowledge
AnswerDesk AI is a deployable AI customer support system for one client at a time. It is not positioned as a multi-tenant SaaS demo.
Core Capabilities
- Website chat widget
- Telegram Bot
- REST API
- Fixed FAQ answers
- RAG knowledge retrieval
- AI playbooks
- Human handoff
- Inbox review
- Action traces
- Reports
- Deployable support workflows
Knowledge Base
Stores factual materials such as product docs, SOPs, PDFs, policies, and FAQs.
AI Playbooks
Define how the AI should handle cases, what it must not promise, and when to hand off to a human.
Hybrid RAG
Combines vector retrieval, keyword retrieval, and reciprocal-rank fusion so answers can match both semantic questions and exact terms such as SKUs, policy names, order words, or product names.
Human Handoff
Can be triggered by:
- Exact keywords
- Semantic classification
- Low confidence
- Sensitive issues
- Missing knowledge
- Customer request
When configured, handoff can emit a webhook event to the client's system.
Language Support
The system supports English and Chinese data isolation. Admin language controls which knowledge, FAQ, playbooks, conversations, and reports are shown and used.
Model Layer
The model layer is replaceable. The current deployment can use a chosen AI model behind the same product architecture.
# AnswerDesk AI Product Knowledge AnswerDesk AI is a deployable AI customer support system for one client at a time. It is not positioned as a multi-tenant SaaS demo. Core capabilities: website chat widget, Telegram Bot, REST API, fixed FAQ answers, RAG knowledge retrieval, AI playbooks, human handoff, inbox review, action traces, reports, and deployable support workflows. The knowledge base stores factual materials such as product docs, SOPs, PDFs, policies, and FAQs. AI Playbooks define how the AI should handle cases, what it must not promise, and when to hand off to a human. Hybrid RAG combines vector retrieval, keyword retrieval, and reciprocal-rank fusion so answers can match both semantic questions and exact terms such as SKUs, policy names, order words, or product names. Human handoff can be triggered by exact keywords, semantic classification, low confidence, sensitive issues, missing knowledge, or customer request. When configured, handoff can emit a webhook event to the client's system. The system supports English and Chinese data isolation. Admin language controls which knowledge, FAQ, playbooks, conversations, and reports are shown and used. The model layer is replaceable. The current deployment can use a chosen AI model behind the same product architecture.