Smart query understanding

Smart query understanding is Chatevo’s pre-retrieval intelligence layer. It maps to the Intent Analyzer and Query Strategist stages in the three-brain pipeline — deciding what the visitor wants and how to fetch an answer before the Response Synthesizer writes a reply.

What it does

Capability	Description
Intent routing	Chooses RAG, tools, hybrid, or direct response
Query rewriting	Expands vague questions into search-friendly queries
Tool selection	Picks the right API Request or MCP tool and parameters
Skip retrieval	Avoids unnecessary KB searches on greetings or off-topic chat

Without smart query understanding (Free plan), Chatevo still runs RAG when knowledge bases are attached, but does not rewrite queries or intelligently route to tools.

Plan tiers

Tier	Plans	Behavior
None	Free	RAG when KBs attached; no routing or rewrite
Basic	Starter	Intent routing + basic query rewrite; API Request tools
Full	Standard, Pro, Enterprise	Full routing, hybrid docs+tools, advanced rewrite, MCP support

Basic vs full

Basic (Starter) — Handles common patterns: “Where is my order?” → tool; “What’s your refund policy?” → RAG. Rewrites short or ambiguous queries before search.

Full (Standard+) — Adds hybrid paths (search docs and call a tool in one turn), better multi-intent handling, and tighter coordination when multiple knowledge bases or tools are attached.

Example flow

Visitor: “Do you have any deals on laptops this week?”
Intent Analyzer — classifies as product lookup needing live inventory.
Query Strategist (full) — calls get_offers tool with category laptops; skips generic FAQ retrieval.
Response Synthesizer — replies with tool JSON; carousel renders from result_transform.

If the tool fails, full tier may fall back to RAG for static promo copy in the knowledge base.

Tuning behavior

Smart query understanding respects assistant configuration:

Attached knowledge bases and tools
RAG settings (threshold, max results)
Master prompt escalation rules