Dashboard

Monitor calls, analyze performance, and manage connections

Overview

The Telepath Dashboard is your central hub for managing connections and monitoring voice AI performance. Access it at dashboard.telepathvoice.com. It provides real-time call data, historical analytics, and tools for diagnosing issues without any additional instrumentation.

Dashboard Features

Connections

The Connections page lists all your Telepath SIP connections. For each connection you can view:

• Status (Active, Inactive, or Error)
• SIP username and configured AI provider
• Total call count and most recent call timestamp
• Average latency metrics for the last 7 days

Available actions on each connection card:

• Edit — update AI provider settings or SIP credentials
• Rotate Password — generate a new SIP password
• View Calls — jump to a filtered call log for this connection
• Delete — permanently remove the connection

Call Log

The Call Log shows every call handled by Telepath with the following details per entry:

• Call ID — unique identifier for the call
• Timestamp — when the call started (UTC)
• Duration — total call length in seconds
• Status — Completed, Failed, or Dropped
• Carrier Lag — time from call arrival to RTP media start
• AI Latency — average time for the AI agent to respond
• Gateway Processing — Telepath's internal processing overhead

Filtering options: Filter by connection, date range, status, minimum duration, or latency thresholds.

Sorting options: Sort by timestamp, duration, carrier lag, AI latency, or gateway processing time.

Call Details (Per-Call Inspection)

Clicking any call in the log opens the full call detail view, which includes:

• Overview: call ID, caller ID, connection used, carrier, timestamps
• Metrics table: carrier lag, AI latency, gateway processing, end-to-end latency, packet loss percentage
• Turn-by-turn analysis: each conversation exchange with individual response latencies
• SIP Trace: full SIP signaling log for debugging authentication and routing issues
• Audio Quality: MOS score estimate, jitter, and packet loss over the call timeline

Analytics

The Analytics section aggregates call data across all your connections:

• Overview stats: total calls, total minutes, average duration, success rate
• Trends: daily/weekly call volume and latency over time
• Provider comparison: side-by-side latency and success rate for each AI provider
• Geographic insights: call volume and quality metrics broken down by caller region

Settings

The Settings section contains:

• Account settings: display name, email, password
• API keys: create, view, and revoke API keys for programmatic access
• Billing: current plan, usage, invoices, and payment method
• Data retention: configure how long call logs and metrics are stored

Key Metrics Explained

Carrier Lag

Carrier Lag measures the time from when a call arrives at Telepath's SIP endpoint to when the first RTP media packet is received. This reflects your carrier's performance.

• Ideal: 50–100 ms
• Acceptable: 100–200 ms
• Investigate: >200 ms consistently

AI Latency

AI Latency measures the average time between Telepath detecting end-of-speech and receiving the first audio frame from your AI provider.

• OpenAI Realtime: typically 100–300 ms
• ElevenLabs: typically 200–500 ms
• Custom WebSocket: depends on your implementation

Gateway Processing

Gateway Processing is the time Telepath itself spends on transcoding, VAD analysis, and internal routing. This should be minimal.

• Excellent: <20 ms
• Normal: 10–50 ms
• Investigate: >50 ms

End-to-End Latency

End-to-End Latency is the total perceived delay from when a caller finishes speaking to when they hear the AI agent's response.

Formula: Carrier Lag + AI Latency + Gateway Processing

• Target: <500 ms for natural conversation feel
• Acceptable: 500–800 ms
• Degraded: >800 ms (users notice the delay)

Monitoring Best Practices

Daily

• Check the Call Log for any failed or dropped calls from the previous day
• Review average AI latency to catch AI provider degradation early
• Confirm all connections are showing an Active status

Weekly

• Review the Analytics trends view for week-over-week call volume changes
• Compare AI latency across different connections or providers
• Check for any recurring error patterns in the call log
• Review carrier lag trends to identify carrier performance regressions

Monthly

• Export call data and review it alongside business KPIs
• Review your billing dashboard to ensure usage is within budget
• Rotate API keys and SIP passwords as part of regular security hygiene
• Evaluate whether your current AI provider is still the best fit based on performance data

Setting Alerts

Configure alerts in Settings → Alerts to get notified when:

• End-to-end latency exceeds a threshold (e.g., 800 ms)
• Call failure rate exceeds a percentage (e.g., 5%)
• A connection's status changes to Error or Inactive
• Call volume drops significantly below the normal baseline (potential carrier outage)

Troubleshooting via Dashboard

Call Dropped Immediately

1. Open the call in the Call Log and view the SIP Trace
2. Look for 4xx or 5xx SIP error codes in the trace
3. A 401 Unauthorized indicates an authentication failure — check SIP credentials
4. A 404 Not Found indicates the origination URI is incorrect
5. A 503 Service Unavailable may indicate a temporary issue with your AI provider

Poor Audio Quality

1. Check the call detail's Audio Quality section for high jitter or packet loss
2. If packet loss >2%, the issue is likely network-related between carrier and Telepath
3. Check the codec used — G.711 at poor network conditions degrades more than G.722
4. Review Carrier Lag — spikes in lag often correlate with audio artifacts
5. Consider switching your carrier to use TLS transport to reduce interference

High Latency

Use the per-call metrics to identify which component is responsible:

• If Carrier Lag is the culprit — contact your carrier or switch providers
• If AI Latency is the culprit — check your AI provider's status page or upgrade your plan
• If Gateway Processing is the culprit — contact Telepath support

Silent Calls

1. Check if the call shows a Completed status but zero AI latency readings
2. This usually means the AI agent connected but returned no audio
3. Review your AI provider configuration — check that the API key is valid
4. For custom WebSocket endpoints, check your server logs for connection errors
5. Review the turn-by-turn analysis to see if any turns were processed

Exporting Data

View Call History

From the Call Log, use the Export button to download call history as a CSV file. You can apply date range and connection filters before exporting.

Accessing Raw Data via API

For programmatic access to call data, use the Telepath REST API:

Tips & Tricks

Custom Dashboard URL: Bookmark https://dashboard.telepathvoice.com/calls?connection={id} to jump directly to a specific connection's call log.

Real-time Monitoring: The Call Log auto-refreshes every 30 seconds. Keep it open in a browser tab during production launches to catch issues immediately.

Comparison Analysis: Use the Analytics → Provider Comparison view to compare AI providers across identical time periods when making provider decisions.

Capacity Planning: Review the Geographic Insights section to understand where your callers are located and plan regional infrastructure accordingly.

API Reference