Analytics Endpoints

Endpoints for retrieving call analytics, performance metrics, and error logs.

Get Analytics Summary

GET /api/v1/analytics/summary

Get aggregated analytics for the current tenant.

Query Parameters:

Parameter	Type	Description
`period`	string	Time period: `1h`, `24h`, `7d`, `30d` (default: `24h`)
`started_after`	datetime	Filter calls started after this time
`started_before`	datetime	Filter calls started before this time

Response:

{
  "total_calls": 1247,
  "completed_calls": 1180,
  "failed_calls": 67,
  "active_calls": 3,
  "avg_duration_seconds": 145.2,
  "success_rate": 94.6,
  "total_tokens": 125000,
  "avg_stt_latency_ms": 120.5,
  "avg_llm_latency_ms": 280.3,
  "avg_tts_latency_ms": 85.2
}

Field	Description
`total_calls`	Total calls in the period
`completed_calls`	Successfully completed calls
`failed_calls`	Failed calls
`active_calls`	Currently active calls
`avg_duration_seconds`	Average call duration
`success_rate`	Percentage of successful calls
`total_tokens`	Total LLM tokens consumed
`avg_stt_latency_ms`	Average speech-to-text latency
`avg_llm_latency_ms`	Average language model latency
`avg_tts_latency_ms`	Average text-to-speech latency

Get Call Volume

GET /api/v1/analytics/call-volume

Get call volume time series data for charting.

Query Parameters:

Parameter	Type	Description
`period`	string	Time period: `1h`, `24h`, `7d`, `30d`
`interval`	string	Bucket size: `hour`, `day`, `week` (default: `hour`)
`started_after`	datetime	Filter by start time
`started_before`	datetime	Filter by end time

Response:

{
  "interval": "hour",
  "data": [
    {
      "timestamp": "2026-01-17T10:00:00Z",
      "total": 45,
      "completed": 42,
      "failed": 3
    },
    {
      "timestamp": "2026-01-17T11:00:00Z",
      "total": 52,
      "completed": 50,
      "failed": 2
    }
  ]
}

Get Workflow Stats

GET /api/v1/analytics/workflow-stats

Get statistics grouped by workflow.

Query Parameters:

Parameter	Type	Description
`period`	string	Time period
`limit`	integer	Max workflows to return (default: 10, max: 50)
`started_after`	datetime	Filter by start time
`started_before`	datetime	Filter by end time

Response:

[
  {
    "workflow_id": "uuid",
    "workflow_name": "Daily Health Check",
    "workflow_slug": "daily-call",
    "total_calls": 500,
    "completed_calls": 475,
    "failed_calls": 25,
    "success_rate": 95.0,
    "avg_duration_seconds": 180.5
  },
  {
    "workflow_id": "uuid",
    "workflow_name": "Onboarding Flow",
    "workflow_slug": "onboarding",
    "total_calls": 150,
    "completed_calls": 142,
    "failed_calls": 8,
    "success_rate": 94.7,
    "avg_duration_seconds": 240.2
  }
]

Get Provider Latency

GET /api/v1/analytics/provider-latency

Get latency statistics by provider type.

Query Parameters:

Parameter	Type	Description
`period`	string	Time period
`started_after`	datetime	Filter by start time
`started_before`	datetime	Filter by end time

Response:

[
  {
    "provider_type": "stt",
    "provider_name": "deepgram",
    "avg_latency_ms": 120.5,
    "p50_latency_ms": 108.4,
    "p95_latency_ms": 180.7,
    "p99_latency_ms": 241.0,
    "call_count": 1247
  },
  {
    "provider_type": "llm",
    "provider_name": "openai",
    "avg_latency_ms": 280.3,
    "p50_latency_ms": 252.2,
    "p95_latency_ms": 420.4,
    "p99_latency_ms": 560.6,
    "call_count": 1247
  },
  {
    "provider_type": "tts",
    "provider_name": "cartesia",
    "avg_latency_ms": 85.2,
    "p50_latency_ms": 76.6,
    "p95_latency_ms": 127.8,
    "p99_latency_ms": 170.4,
    "call_count": 1247
  }
]

Field	Description
`provider_type`	Type: `stt`, `llm`, or `tts`
`provider_name`	Provider name (e.g., deepgram, openai)
`avg_latency_ms`	Average latency in milliseconds
`p50_latency_ms`	50th percentile (median) latency
`p95_latency_ms`	95th percentile latency
`p99_latency_ms`	99th percentile latency
`call_count`	Number of calls with this provider

Get Error Logs

GET /api/v1/analytics/errors

Get error logs from calls.

Query Parameters:

Parameter	Type	Description
`period`	string	Time period (default: `24h`)
`service`	string	Filter by service: `stt`, `llm`, `tts`, `workflow`, `tool`
`level`	string	Filter by level: `error`, `warning`
`offset`	integer	Pagination offset (default: 0)
`limit`	integer	Max records (default: 50, max: 200)
`started_after`	datetime	Filter by start time
`started_before`	datetime	Filter by end time

Response:

{
  "items": [
    {
      "id": "call-uuid-0",
      "call_id": "uuid",
      "timestamp": "2026-01-17T10:30:00Z",
      "service": "stt",
      "level": "error",
      "message": "Connection timeout to Deepgram",
      "details": {
        "timeout_ms": 5000,
        "retry_count": 3
      },
      "agent": "meal-agent"
    },
    {
      "id": "call-uuid-1",
      "call_id": "uuid",
      "timestamp": "2026-01-17T10:25:00Z",
      "service": "llm",
      "level": "error",
      "message": "Rate limit exceeded",
      "details": {
        "provider": "openai",
        "retry_after": 60
      },
      "agent": "greeter-agent"
    }
  ],
  "total": 67
}

Error Services

Service	Description
`stt`	Speech-to-text provider errors
`llm`	Language model provider errors
`tts`	Text-to-speech provider errors
`workflow`	Workflow execution errors
`tool`	Tool execution errors

Error Levels

Level	Description
`error`	Critical errors that affected call
`warning`	Non-critical issues

Get Active Calls Count

GET /api/v1/analytics/active-calls

Get the count of currently active calls. Useful for real-time monitoring dashboards.

Response:

{
  "count": 3
}

Time Periods

All analytics endpoints support the period parameter:

Period	Description
`1h`	Last hour
`24h`	Last 24 hours (default)
`7d`	Last 7 days
`30d`	Last 30 days

Alternatively, use started_after and started_before for custom date ranges:

GET /api/v1/analytics/summary?started_after=2026-01-01T00:00:00Z&started_before=2026-01-15T23:59:59Z

Dashboard Refresh Rates

Recommended refresh intervals for dashboard polling:

Endpoint	Refresh Rate
`/analytics/summary`	30 seconds
`/analytics/active-calls`	10 seconds
`/analytics/call-volume`	60 seconds
`/analytics/workflow-stats`	60 seconds
`/analytics/provider-latency`	60 seconds
`/analytics/errors`	30 seconds

For real-time updates, use WebSocket endpoints instead.

Analytics Endpoints

On this page