Architecture
Smart Routing Engine

Smart Routing Engine

Smart Routing uses an internal Small Language Model (SLM) to classify query intent and automatically select the optimal LLM for each query.

What Smart Routing Does

When Smart Routing is enabled, the SLM analyzes each query and considers:

  • Query complexity — Simple questions route to faster, cheaper models; complex reasoning routes to flagship models
  • Model capabilities — Matches query requirements (coding, analysis, creative) to model strengths
  • Cost efficiency — Selects the most cost-effective model that can handle the query
  • Latency requirements — Prioritizes faster models for time-sensitive queries

How It Works

  1. The SLM receives the query and classifies its intent
  2. The classification is matched against available models
  3. The model with the best capability-to-cost ratio for that intent is selected
  4. If no clear match, the organization's default model is used

Info: Smart Routing is transparent — users and agents always send queries to the same endpoint. The routing decision is visible in the Query Log.

Enabling Smart Routing

  1. Navigate to SuperAdmin > Model Config.
  2. Toggle Smart Routing Enabled on.
  3. Smart Routing will be used for queries that do not match a specific routing rule.

Smart Routing vs. Routing Rules

  • Routing Rules — Explicit rules you create to route specific query types to specific models
  • Smart Routing — Automatic model selection for queries that do not match any routing rule
  • Both can be used together — routing rules take priority, Smart Routing handles the rest