Cost-aware agent dispatch

The dispatcher inspects a query and routes it to a cheap-fast or expensive-thorough agent - same UX, lower cost

Cost-aware agent dispatch The dispatcher inspects a query and routes it to a cheap-fast or expensive-thorough agent - same UX, lower cost 01 / Client 02 / Cheap-fast agent 03 / Dispatcher 04 / Thorough agent Inspect + route Serve User query · one interface · Client › Inspect + route User query one interface Classify · cost + complexity · Dispatcher › Inspect + route · budget rule Classify cost + complexity budget rule Cheap-fast · small model · Cheap-fast agent › Serve · low cost Cheap-fast small model low cost Thorough · large model + tools · Thorough agent › Serve · high cost Thorough large model + tools high cost Answer · identical UX · Client › Serve Answer identical UX prompt simple complex Legend User UI Agent logic Policy Tool action Context / trace

Inspect

  • • Estimate token cost and task complexity
  • • Score against a budget policy
  • • No model call needed to route

Route by cost

  • • Simple queries take the cheap-fast path
  • • Hard queries earn the expensive-thorough path
  • • Policy is tunable per tenant or budget

Same UX

  • • Caller sees one endpoint and one response shape
  • • Latency and spend drop on the common case
  • • Quality is preserved where it matters

Workflow diagram • Built with Archify • Create yours ↗ • Hover to trace • R route • Click to focus • +/− zoom • M radar • T theme • E export