Local AI Advisor
realwllama local inference with explicit dictionary fallback.
Purpose
AI Advisor can load a local GGUF model in-browser using wllama, stream responses, and fall back to curated dictionary mode when no model is loaded.
When to Use
- When interpreting tasks
- When generating field instructions
- When converting recommendations into tasks
- When offline and no human expert is available
Best First Action
If bandwidth and power allow, load the smallest model first. Otherwise use dictionary mode knowingly.
Content
Purpose
AI Advisor can load a local GGUF model in-browser using wllama, stream responses, and fall back to curated dictionary mode when no model is loaded.
When to Use
- When interpreting tasks
- When generating field instructions
- When converting recommendations into tasks
- When offline and no human expert is available
Best First Action
If bandwidth and power allow, load the smallest model first. Otherwise use dictionary mode knowingly.
Best Practices
- Prefer operational mode for direct action.
- Use Socratic mode for teaching.
- Use JSON mode for future automation handoff.
- Never treat AI advice as field verification.
Common Mistakes
- Loading a 1GB model on weak battery.
- Copying AI output into field work without checking maps/sensors.
- Assuming dictionary fallback is a live model.
Next Steps
- Convert good guidance into a task.
- Log field results in History.
- Use Knowledge Base pages as grounding context.
Best Practices
- Prefer operational mode for direct action.
- Use Socratic mode for teaching.
- Use JSON mode for future automation handoff.
- Never treat AI advice as field verification.
Common Mistakes
- Loading a 1GB model on weak battery.
- Copying AI output into field work without checking maps/sensors.
- Assuming dictionary fallback is a live model.
Next Steps
- Convert good guidance into a task.
- Log field results in History.
- Use Knowledge Base pages as grounding context.