AI Routing & Eco Mode
The Oakden Router is the intelligence layer that automatically directs every AI request to the most efficient provider. It balances performance, cost, and privacy based on your preferences.
How Routing Works
When you make an AI request (a diagnosis, a chat message, a document scan), the router evaluates:
- Request complexity — Simple questions can be handled locally. Complex reasoning or vision tasks may benefit from cloud models.
- Hardware capabilities — What local AI models are available on your hardware and whether they can handle the request.
- Your routing preference — The mode you have selected (Eco, Balanced, or Performance).
- Current load — If local hardware is busy, the router may queue the request or route to cloud.
The routing decision happens in milliseconds and is completely transparent. You can see where each request was processed in the usage dashboard.
Eco Mode
Eco Mode maximizes local inference. Every request that can be handled by a local model stays local. Cloud is used only when absolutely necessary (for example, a vision task that requires a model not available locally).
Benefits of Eco Mode:
- Lowest cost — Local inference costs only electricity
- Maximum privacy — Almost all data stays on your hardware
- Fast for routine tasks — Local models respond in milliseconds for common queries
Eco Mode is ideal for daily operations, routine AI chat, and environments where privacy is the top priority.
Balanced Mode
Balanced Mode uses a mix of local and cloud based on the best fit for each request. Simple tasks run locally. Complex tasks route to cloud for better results. This is the default mode and works well for most users.
Benefits of Balanced Mode:
- Best quality-to-cost ratio — You get cloud quality when it matters and local efficiency for everything else
- Automatic optimization — The router learns which tasks benefit most from cloud processing
- Good privacy — Most routine data stays local
Performance Mode
Performance Mode routes to the most capable model available regardless of cost. This typically means cloud providers with the latest large language models. Use this mode when you need the highest-quality AI output for important tasks like client presentations, complex estimates, or critical document analysis.
Real-Time Dashboard
The usage dashboard shows:
- Routing split — What percentage of requests went local versus cloud
- Cost tracking — Actual cost per request and cumulative spending
- Model usage — Which models are being used and how often
- Savings estimate — How much you saved by using local inference versus cloud-only pricing
Changing Your Mode
You can switch between modes anytime in Settings > AI Routing. The change takes effect immediately for all subsequent requests. There is no penalty or delay for switching.
Token Budget
For organizations that want to control cloud AI spending, ProAI Builder supports a token budget. Set a monthly cap on cloud AI tokens, and when the budget is reached, the system gracefully falls back to local models only. This prevents unexpected cloud bills while keeping the AI functional.