Routing requests across models
Business
Routing requests across models
- A Test Suite for Precedence Rules: Cases That Catch Drift
- A/B Testing Model Routing Rules Without Breaking Production
- Combining Two Signals Without Double-Counting: A Lift Test
- How to Tag and Route Inbound Questions by Topic Without an LLM
- Routing Requests by Cost and Quality: Rules You Can Audit
- Small Model First, Larger on Retry: Routing by Tier
- Ticket Signals That Predict a First-Pass Failure
- When Two Routing Rules Match the Same Request
- A Holdout Group for Router Changes: How Long and How Big
- A Routing Ladder for Three Tiers: When the Second Retry Climbs
- A Scorecard for Choosing Between Two Router Signals
- Backtesting a Router Signal on Historical Tickets Without Leakage
- Classifying Inbound Messages by Intent Before the Agent Answers
- How to Backtest a New Router Signal Without Leakage
- Keeping a Routing Decision Log a Reviewer Can Follow Later
- Ledger Fields That Let You Compare Routing Policies
- Logging a Request That Climbs: Replay and Cost Fields
- Logging the Router Decision So You Can Recompute Lift Later
- Logging the Routing Decision Fields That Let You Replay a Request
- Overlap Shapes in Rule Sets: Exact, Prefix, Range and Negation
- Reading a Lift Estimate When the Sample Is Small
- Routing by Failure Type: Prompt Fix or Model Climb?
- Routing Requests to the Cheapest Model That Still Passes
- Rule Precedence When Two Routing Rules Match
- The Holdout Group for Router Changes: How Long and How Big
- When a Router Signal Goes Quiet: Drift Checks and a Review Date
- When a Router Signal Stops Working: Drift Checks
- When to Use a Smaller Model for the Easy Questions