Textual Signals
What users type or say carries emotional weight. Nefesh accepts sentiment and urgency scores extracted from text, adding a linguistic dimension to the stress assessment.
Accepted Fields
| Field | Type | Range | Description |
|---|---|---|---|
sentiment | float | -1.0 to 1.0 | Text sentiment score. -1.0 = very negative, 0 = neutral, 1.0 = very positive |
urgency | string | low, medium, high, critical | Urgency classification from message content |
How It Works
You run sentiment analysis on the user's text (using any NLP pipeline) and send the result. Nefesh uses this alongside biometric signals to understand context — for example, negative sentiment combined with elevated heart rate is a stronger stress indicator than either signal alone.
Trigger Memory Integration
For deeper textual analysis, send user_message and ai_response alongside your signals. Nefesh will automatically extract psychological topics and build a per-user trigger profile over time. See the API Reference for details.
Example Payload
{
"session_id": "sess_abc123",
"timestamp": "2026-03-30T14:30:00Z",
"sentiment": -0.6,
"urgency": "high",
"heart_rate": 88,
"user_message": "I can't handle this deadline anymore",
"ai_response": "I understand that feels overwhelming. Let's break it down."
}Note: sentiment is a float, not a string. Use a value between -1.0 and 1.0.