Analytics & Observability

Workspace analytics, call management, recording access, and the external-write review queue.

Workspace-level analytics, call management, recording access, and the external-write review queue.

Analytics

• Usage summary: Event counts by type with daily breakdown
• Call statistics: Volume and duration metrics
• Event breakdown: Distribution by type and source
• Data quality: Confidence distribution and review pipeline stats

Call Intelligence

Per-call intelligence summaries are computed from in-memory session state at call end and persisted for analytics use. Each record includes a composite quality score (0-100), emotion summary, risk assessment, latency metrics, conversation shape, tool performance, safety matches, and operator data. See Voice Agent, Call Intelligence Persistence for field details and scoring rules.

Call Intelligence Analytics (5 Endpoints)

Five endpoints aggregate call intelligence data for dashboard visualizations.

Common parameters (all endpoints):

Parameter	Type	Default	Description
days	integer (1-90)	30	Lookback window if no explicit dates
date_from	date	-	Start of range (inclusive)
date_to	date	-	End of range (inclusive)
interval	1h / 1d / 1w	1d	Time bucket for trend data
service_id	string	-	Filter to a specific service

Endpoints:

Endpoint	Response Shape
GET /v1/{workspace_id}/analytics/call-quality	summary (avg/p50/p95 quality, total calls, escalation rate, avg duration) + trend (per-bucket avg quality, call count, escalation count) + quality_distribution (excellent/good/fair/poor counts)
GET /v1/{workspace_id}/analytics/emotion-trends	emotion_distribution (emotion name + count, sorted by frequency) + trend (per-bucket call count, avg valence, avg arousal)
GET /v1/{workspace_id}/analytics/latency	summary (per-component p50/p95/p99) + trend (per-bucket component percentiles)
GET /v1/{workspace_id}/analytics/tool-performance	summary (total calls, overall failure rate) + tools (per-tool invocations, failures, failure rate, avg ms) + trend (per-bucket tool calls, failure rate)
GET /v1/{workspace_id}/analytics/operator-performance	summary (escalation count/rate, avg connect time, avg quality for escalated vs non-escalated) + trend (per-bucket escalation counts)

All endpoints require viewer+ permissions and are workspace-scoped. Quality distribution tiers: excellent (90-100), good (70-89), fair (50-69), poor (0-49).

Calls & Recordings

• Call listing: Filter by date, status, and duration. Each call includes quality score and final state from call intelligence. Supports continuation_token for cursor-based pagination alongside offset.
• Call detail: Full call detail with turns, timeline, and playback data (proxied from the voice agent for rich conversation reconstruction)
• Call intelligence: Per-call intelligence profile with emotion summary, risk summary, latency breakdown, and tool performance
• Active call intelligence: Live intelligence overlay for all active calls (current emotion, turn count, escalation status)
• Phone volume: Call volume per phone number over a time window
• Recordings: Time-limited signed URLs for audio playback, plus recording metadata

Review Queue

The review queue holds external write proposals awaiting a human decision: list pending proposals, inspect a proposal's payload, and approve or reject it. See Approval-Gated Integration Writes for the full lifecycle and the Review Queue API reference for the endpoint schemas.

Population Health Analytics

A layered epidemiology pipeline materializes six per-workspace analytics output tables for downstream risk stratification, dashboard visualizations, and explainable cohort analysis:

• patient_condition_fact: exploded patient × condition fact with onset, age at onset, normalized code system, and cohort bucket.
• disease_trends_monthly: incident counts by code × month × cohort.
• disease_prevalence_current: active-condition prevalence with Wilson 95% confidence intervals, which stay reliable for small samples.
• disease_cohort_drivers: Standardized Prevalence Ratio (SPR - the ratio of a cohort's observed condition prevalence to the prevalence expected from the overall population) per (code, cohort) with Byar 95% confidence intervals. The canonical measure of whether a cohort over- or under-indexes for a condition.
• patient_condition_burden: per-patient Charlson Comorbidity Index (CCI - a weighted comorbidity score that predicts mortality risk from a patient's condition list; Quan 2005 ICD-10 mapping) and age-adjusted CCI (Charlson 1994). Feature-ready for risk models.
• condition_cooccurrence_pairs: association statistics for condition pairs - Jaccard similarity, lift, pointwise mutual information (PMI - how much more often two conditions co-occur than chance would predict), and chi-square.

All association and ratio tables emit a min_support_met flag at the standard small-cell suppression threshold (observed ≥ 5). Filter on it to avoid reporting unstable rates.

Methodology follows the standard epidemiology literature: Wilson (1927), Breslow & Day (1987), Charlson et al. (1987, 1994), Quan et al. (2005).

Platform Health Monitoring

Two endpoints provide real-time platform health visibility without querying the analytics warehouse:

GET /v1/{workspace_id}/sensorium/connector-health. Per-source event health for every data source writing into the workspace.

Each source entry includes:

Field	Description
events_last_hour	Events ingested in the last hour
events_last_24h	Events ingested in the last 24 hours (rolling, not calendar-day)
events_per_minute	Mean events per minute over the last hour
last_ingested_at	Timestamp of last ingested event
freshness	fresh (<5m), stale (5-60m), quiet (>60m), never (no events in 24h)

GET /v1/{workspace_id}/sensorium/loop-latency. End-to-end latency from data ingestion to agent action.

Field	Description
overall_p50_seconds	Median sense-to-act latency in seconds across the window; null when no pairs
overall_count	Total sensing-to-action pairs in the window
window_hours	Echoed window size
truncated	true when the underlying query hit the row cap, so the median is computed over a truncated sample
as_of	Server timestamp of the query (UTC)
sparkline[]	Up to 168 hourly buckets, each with hour (UTC hour bucket), count, and p50_seconds (null if no pairs that hour)

Query parameter: window_hours (1-168, default 24) controls the lookback window.

Both endpoints are workspace-scoped and rate-limited under the standard read policy. Permissions: viewer+.

Voice Judge

Audio-native quality evaluation that scores every call across ten dimensions directly from the stereo recording. Unlike post-interaction quality scoring (which analyzes structured session data), the Voice Judge operates on the full audio and produces per-dimension scores with structured evidence for UI drill-down. The ten dimensions are also available as aggregate built-in metrics in the metric store, so workspace-level trends flow through the standard dashboard and alerting infrastructure.

Voice Judge

Pipeline Health

Read-only endpoints under /v1/{workspace_id}/pipeline/ provide real-time connector runner observability for the pipeline dashboard: composite pipeline status, per-source health and history, outbound sync summaries and logs, entity resolution metrics, review pipeline metrics, and event throughput. The endpoints degrade gracefully when the connector runner is unavailable - stored metrics stay visible. See Connector Runner, Pipeline Observability for the full endpoint reference.

Command Center

GET /v1/{workspace_id}/command-center. Single-pane workspace health dashboard.

Aggregates metrics from five domains concurrently, each failing independently:

Section	Response Model	Key Fields
voice	VoiceMetrics	active_calls, active_escalated, calls_today, avg_quality_score_today, escalation_rate_today
pipeline	PipelineMetrics	sources_total, sources_healthy, sources_degraded, sources_failing, events_last_hour, outbound_pending, outbound_failed, connector_status
data_quality	DataQualityMetrics	pending_reviews, approval_rate_7d, avg_confidence, entities_total, entities_merged_7d
identity	IdentityMetrics	active_api_keys, active_sessions, failed_auths_1h, locked_accounts, mfa_coverage_pct
alerts	CommandCenterAlert[]	level (warning/critical), code, message, section

Additional response fields: degraded_sections (list of section names that failed to load), generated_at, cached.

Caching: Responses may be served from a short-lived cache; the cached field indicates when this happened. Alerts: Derived from 7 threshold checks on the aggregated metrics. Permissions: viewer+.

Percentile Analytics

GET /v1/{workspace_id}/analytics/calls/advanced. Percentile-based call metrics.

Parameter	Type	Default	Description
days	integer (1-90)	30	Lookback window
date_from / date_to	date	-	Explicit date range
interval	1h / 1d / 1w	1d	Time bucket
service_id	string	-	Filter by service
direction	string	-	Filter by inbound or outbound

Returns: summary (p50/p95/p99 for duration + quality), trend (per-bucket with p95 latency), by_service breakdown, by_direction breakdown.

GET /v1/{workspace_id}/analytics/calls/comparison. Period-over-period comparison.

Parameter	Type	Required	Description
current_from / current_to	date	Yes	Current period
previous_from / previous_to	date	Yes	Previous period
service_id	string	No	Filter by service

Returns: current metrics, previous metrics, delta (absolute + percentage change for each KPI).

Entity Intelligence & World Sync

Entity-level intelligence endpoints (relationship graph, provenance, lineage, duplicate detection, entity search) and the outbound world sync events endpoint are documented with the rest of the world model API.

Data & World Model

Real-Time Event Stream

GET /v1/{workspace_id}/events/stream. Server-Sent Events for real-time workspace updates.

Subscribes to the workspace event channel and streams events as they occur:

Event Type	When It Fires
call.started	New call begins
call.ended	Call completes
call.escalated	Operator escalation triggered
pipeline.sync_completed	Data source poll finished
pipeline.error	Pipeline error detected
review.submitted	Review queue item processed
alert	Alert threshold crossed

Protocol details:

• Media type: text/event-stream
• Heartbeat: Comment every 30s to keep connection alive
• Reconnection: Send Last-Event-ID header on reconnect for replay of recent events
• Backpressure: Oldest events dropped when the event buffer is full
• Falls back to error event if the event system is unavailable

Permissions: viewer+.

Dashboards

Workspace-scoped analytics dashboards that can be rendered in the Developer Console or embedded in external applications.

Dashboards are managed through CRUD endpoints under /v1/{workspace_id}/dashboards/. Each dashboard has a slug (URL-friendly identifier), title, description, and chart configuration. The list endpoint returns all dashboards for the workspace. The get-by-slug endpoint returns the full dashboard definition including data queries and chart layout.

The @amigo-ai/dashboards package provides an embeddable React component (DashboardEmbed) that renders a dashboard given a workspace ID and slug. The component supports two authentication modes: explicit API token or backend-for-frontend proxy (where the hosting application handles authentication and the component omits the token).

API Reference

• Analytics
• Calls
• Recordings
• Review Queue
• Command Center
• Events