Analyse Conversations.
Surface What Matters.
InnerLayer detects psychological state, behavioural patterns, and intent signals in real time — transforming raw conversation into structured, actionable intelligence.
Authentication
All requests require an API key passed as a request header. Keys are scoped to your organisation and deployment context.
# Include with every request x-api-key: il_live_xxxxxxxxxxxxxxxxxxxx
Getting your key: API keys are issued on request. Visit innerlayer.ai or contact api@innerlayer.ai to request access.
Analyse
Submit a conversation as raw text or a structured transcript. Returns a full behavioural analysis including psychological state, intent, risk signal, and recommended response type.
Body Parameters
The conversation to analyse. Accepts plain text or a full transcript. Maximum 50,000 characters.
Processing mode. sync returns immediately (default). batch accepts file upload for post-call analysis.
Deployment context. Shapes how behavioural markers are weighted and how commercial_translation is phrased. Defaults to general.
Structured turn-by-turn format. Each object requires speaker and text fields. Use instead of conversation for multi-speaker transcripts.
Request Examples
Response
{
"session_id": "il_9f3a2c7e4d",
"psychological_state": "Controlled Withdrawal",
"persona": "LINGUISTIC WITHDRAWAL",
"intent": "disengaging without explicit objection",
"risk_signal": "high",
"behavioural_markers": [
"shortened responses",
"hedging language",
"internal escalation deflection",
"social proof seeking"
],
"response_type": "ANCHOR",
"commercial_translation": "Deal at risk of silent loss — intervene now",
"dominant_pattern": "Agreeable Deflection",
"session_label": "Breakthrough",
"follow_through_delta": {
"from": 50,
"to": 92,
"direction": "positive"
},
"turns_detected": {
"flagged": 6,
"total": 9
},
"confidence": 0.87,
"processed_at": "2026-04-04T11:18:00Z"
}Response Fields
| Field | Type | Description |
|---|---|---|
| session_id | string | Unique identifier for this analysis session. |
| psychological_state | string | Human-readable emotional and behavioural state label. |
| persona | string | Named behavioural archetype from InnerLayer's taxonomy (e.g. LINGUISTIC WITHDRAWAL, THE DOER, FEAR OF BECOMING). |
| intent | string | Detected underlying intent, distinct from stated intent. |
| risk_signal | enum | low · medium · high · critical |
| behavioural_markers | array | Detected strings from InnerLayer's 861-marker behavioural taxonomy. |
| response_type | enum | Recommended conversational action. See Response Types section. |
| commercial_translation | string | Plain-language operational directive adapted to your deployment context. |
| dominant_pattern | string | The primary behavioural pattern detected across the conversation. |
| session_label | string | Top-level session classification (e.g. Breakthrough, At Risk, Stalled). |
| follow_through_delta | object | Predicted follow-through shift. Includes from, to, and direction. |
| confidence | float | Model confidence score (0.0–1.0). |
Batch
Submit a TXT or CSV file containing one or more transcripts for asynchronous post-call analysis. Results are returned as a structured report per session.
Transcript file. Accepted formats: .txt, .csv. Maximum 10MB per request.
Deployment context applied to all sessions in the batch.
Response Types
Every analysis returns a response_type — a recommended conversational action adapted to your deployment context.
Conversation is tracking well. Maintain current approach.
Reduce complexity. State the ask clearly and stop.
Re-establish commitment. Risk of silent exit is elevated.
Surface the real objection. Do not move to close yet.
Validate mutual understanding before proceeding.
Reduce pressure. The speaker needs space to decide.
Do not respond yet. Let the behavioural pattern complete.
All types are adapted to the context parameter in each request.
Behavioural Personas
InnerLayer classifies speakers into named behavioural archetypes. Each persona reflects a distinct psychological state pattern detected across 31,000+ hours of clinical observation.
Disengaging through language reduction. Responses shorten, hedging increases, commitment language disappears.
Action-oriented, impatient with process. High follow-through intent. Respond with directness and decisional clarity.
Ambivalence about change. High surface engagement masking deep resistance. Requires anchoring to desired future state.
Agrees readily but defers consistently. High risk of silent non-conversion. Intervene before the window closes.