[FEATURE] Add example with routing requests based on similarity #3
Description
Problem Statement
Using an LLM as classifier for routing is inefficient.
Proposed Solution
Route requests based on similarity. Search vectordb for existing items with high similarity to link as a duplicate. Search vectordb for the closest category cluster.
openai embeddings
Pattern Category
Which category does this belong to?
- Generative Parsing (structuring data)
- Semantic Routing (directing flow)
- Grounded Pipelines (contextual integrity)
- New Pattern Category
- Adapters/Infrastructure
Use Case
Route tickets based on similarity, duplicates get linked, new ones get created