🎯

Search Intent Classification

Classify every search query as navigational, transactional, informational, or commercial, and disambiguate ambiguous terms like 'apple' (tech vs fruit). Route queries to the right search strategy before they hit the index.

First step in pipeline — Intent classification happens before expansion, ranking, or rescue. Must be near-zero latency
Disambiguation — 'Apple' → tech vs fruit vs brand. Routes to the right search strategy, no heuristic rules
Vertical-extensible — Add breed names vs product names via LEAP for pet retail. Heuristic rules can't scale

The Problem

Does 'apple' mean MacBook or Honeycrisp? Without intent classification, every query hits the same retrieval strategy.

How LFM Compares

Without classification, every query hits the same retrieval path. Heuristic rules handle simple cases but miss ambiguity. LFM classifies intent and disambiguates at 45ms per query.

What LFM Unlocks

Intent + disambiguation at 45ms, routing each query to the optimal retrieval strategy before it hits the index.

This demo is fine-tuned on sample data. Results improve with your data.