| name | ai-routing |
| description | ARAYA Provider-Agnostic AI Routing — capability-based model selection, cost governance, and explainable routing decisions. |
AI Routing Engine
Agents own expertise. Models provide execution capacity. Route work to the
best execution engine while preserving organizational governance.
Routing Strategies
| Strategy | Optimizes For |
|---|
| FAST | Latency — smallest/fastest model capable of the task |
| BALANCED | Quality vs. cost tradeoff |
| REASONING | Deep analysis — largest context, highest reasoning |
| ECONOMY | Lowest cost model capable of the task |
| ENTERPRISE | Governance-first — respects all constitutional rules |
Cost Classes
| Class | Description |
|---|
| LOW | Most economical — simple tasks, documentation, formatting |
| MEDIUM | Balanced — standard development, testing, review |
| HIGH | Premium reasoning — architecture, security, planning |
| PREMIUM | Maximum capability — mission-critical analysis |
Commands
/araya route "<task>" — recommend agent + provider + model for a task/araya route --explain — show why a routing decision was made/araya provider:list — list registered providers and their models/araya model:list — list models with capabilities and routing classes
Rules
- Capability matching is mandatory — no routing without capability verification
- Routing must be explainable — every decision has documented reasoning
- Cost optimization must not violate quality requirements
- Routing respects organizational technology preferences
- Historical success patterns (trajectories) inform routing decisions
