[arXiv]score: 0.11
SLM-Based Orchestration Gateway Routes Prompts to Heterogeneous AI Backends
June 3, 2026
An edge-deployed small language model classifies user intent and routes requests to appropriate AI backends via a configurable service registry, decoupling virtual world clients from specific model dependencies. The architecture targets extensibility across edge and cloud without embedding model logic into the client.
cs.AIcs.HC
HOW THIS AFFECTS YOU
●
builderThe intent-classification-plus-registry pattern is directly applicable to multi-model API routing in production systems beyond virtual worlds.
●
researcherThe paper formalizes an SLM-as-orchestrator architecture with edge deployment constraints — a concrete reference design for lightweight agent routing research.