Getting Started

Observe Quick Start

Detect emotional shifts and escalation signals — 2 minutes

What Observe Does

The Observe API captures lightweight "salience records" at topic-shift boundaries. Unlike full extraction, Observe focuses on detecting what changed — topic shifts, affect changes, risk signals — and whether escalation is recommended.

Use Observe for real-time monitoring of emotional AI in production. Capture what matters without the overhead of full artifact governance.

Step 1: Get an API Key

Select the Observe scope

Generate an API key from the console. When creating your key, enable the observe scope under the Observe section.

Step 2: Make Your First Call

Capture a salience record

Call POST /v1/observe with the content you want to analyze. Include a subject ID and optional session ID for threading.

curl -X POST https://www.deepadata.com/api/v1/observe \
  -H "Authorization: Bearer dda_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "I have been journaling for three months now and I notice I keep coming back to this feeling of being stuck in my career. Today it shifted though - I realized the stuck feeling is actually fear of failing if I try something new.",
    "subject_id": "user-12345",
    "session_id": "journal-session-789"
  }'

The API returns within ~5 seconds with the salience analysis.

Step 3: Read the Response

Understand the salience record

{
  "success": true,
  "data": {
    "record": {
      "id": "sal_abc123...",
      "trigger": "topic_shift",
      "topic_label": "career anxiety → self-awareness",
      "state_change": "Shifted from feeling stuck to recognizing underlying fear of failure",
      "affect": {
        "valence": "mixed",
        "intensity": 0.72,
        "primary_emotion": "fear"
      },
      "significance": 0.85,
      "escalate_recommended": false,
      "themes": ["career", "self-discovery", "fear", "growth"],
      "risk_markers": []
    },
    "meta": {
      "subject_id": "user-12345",
      "session_id": "journal-session-789",
      "created_at": "2026-02-28T10:30:00.000Z"
    }
  }
}

Key Fields

triggerWhat fired: topic_shift, affect_shift, risk_signal, etc.
affect.valenceEmotional direction: positive, negative, mixed, or neutral
escalate_recommendedBoolean — does this warrant human review?
topic_labelHuman-readable description of the shift
significance0.0 to 1.0 — how emotionally significant is this moment?

Lightweight by design: Observe records capture what changed and why it matters — not the full content. Use them for monitoring, alerting, and retrieval-by-salience.

What's Next?