AI features
AI features, privacy, and human review
Use Yanola Agent, AI Response, Classifier, Extract Info, Detect Mood, and AI-assisted Function drafts responsibly.
Automation buildersAdministratorsBeginnerAvailable
What this guide covers
- Yanola AI features generate drafts or runtime outputs from prompts and workflow context; they can be incorrect and must be bounded by validation and human escalation.
- Prerequisites: A defined business purpose.
- Result: Outputs are useful within the documented schema and uncertain cases reach a human.
- Limit: AI output is probabilistic. It must not be the sole authority for high-impact decisions, and workflow tests execute AI nodes for real.
What this is and when to use it
Yanola AI features generate drafts or runtime outputs from prompts and workflow context; they can be incorrect and must be bounded by validation and human escalation.
| Use it when | Expected outcome |
|---|---|
You want generated text, classification, extraction, mood detection, or an automation draft. | Outputs are useful within the documented schema and uncertain cases reach a human. |
Before you begin
- A defined business purpose.
- Only the minimum customer and business data needed for that purpose.
- A human-review path for sensitive or consequential decisions.
Prerequisites
A defined business purpose. Only the minimum customer and business data needed for that purpose. A human-review path for sensitive or consequential decisions.
Complete the task
- Choose the narrowest AI node for the task.
- Write explicit instructions and constrain categories or extraction fields.
- Avoid placing secrets, API keys, tokens, or unnecessary sensitive data in prompts.
- Test representative and adversarial examples.
- Add conditions and Human Handoff for low-confidence or sensitive outcomes.
- Monitor outputs and revise prompts.
Confirm it worked
- Outputs are useful within the documented schema and uncertain cases reach a human.
Common problems
| Symptom | What to check |
|---|---|
Output is inconsistent | Tighten instructions, categories, examples, and downstream validation. |
Sensitive data appears in a prompt | Remove it and rotate any credential that was exposed. |
Current limitations
AI output is probabilistic. It must not be the sole authority for high-impact decisions, and workflow tests execute AI nodes for real.
Product boundary
AI output is probabilistic. It must not be the sole authority for high-impact decisions, and workflow tests execute AI nodes for real.
Where to go next
Related documentation
Yanola Agent workflow assistantAsk the in-builder AI assistant to draft, inspect, or repair a workflow while keeping review and activation under human control.AI ResponseGenerate an AI response to the message. Exact configuration, outputs, execution behavior, errors, limits, and examples for the AI Response workflow node.ClassifierSort messages into categories automatically. Exact configuration, outputs, execution behavior, errors, limits, and examples for the Classifier workflow node.