Fix common Dasha BlackBox problems quickly. Find your symptom, identify the cause, apply the solution.Documentation Index
Fetch the complete documentation index at: https://docs.blackbox.dasha.ai/llms.txt
Use this file to discover all available pages before exploring further.
Quick Diagnosis
| Symptom | Likely Cause | Solution |
|---|---|---|
| Agent enabled but calls fail | Phone not configured | Agent Not Responding |
| HTTP 401 errors | Invalid/expired API key | Authentication Failures |
| Webhooks not received | URL not accessible | Webhook Failures |
| ”Concurrent limit exceeded” | Too many active calls | Concurrent Limits |
| Tool calls failing | Webhook timeout or schema mismatch | Tool Failures |
| Robotic/choppy voice | TTS provider or network | Voice Quality |
| Agent mishears user | Language or noise settings | ASR Errors |
| Slow/irrelevant responses | Model or prompt issues | LLM Issues |
| Recording not found | Not enabled or still processing | Recording Issues |
Agent Not Responding
Symptoms: Agent shows “Enabled” but calls don’t connect, go to voicemail, or fail immediately.Quick Checks
| Check | How to Verify | Fix |
|---|---|---|
| Agent status | Dashboard → Agent → Status badge | Enable if Disabled |
| Phone number | Agent config → phoneNumber field | Assign E.164 format number (+12025551234) |
| Required fields | System prompt, LLM, TTS, STT configured | Fill missing fields |
Verify via API
Verify via API
Debug Steps
- Test from dashboard — If dashboard test works but live calls fail, issue is phone/integration setup
- Check call logs — Dashboard → Calls → Filter by agent → Review failed calls
- Verify webhooks — Ensure webhook URLs are publicly accessible (not localhost)
Authentication Failures
Symptoms: HTTP 401 Unauthorized, “Invalid API key”, “Token expired”Common Causes
| Error | Cause | Fix |
|---|---|---|
| ”Invalid API key” | Key doesn’t exist or revoked | Generate new key in dashboard |
| ”Token expired” | Web integration token expired | Regenerate via API or dashboard |
| 401 only in production | Using wrong environment key | Check environment variables |
| 401 from specific origins | Origin not whitelisted | Add origin to web integration |
Verify API Key
Verify API Key
Fix Origin Restrictions
Fix Origin Restrictions
For web integrations, add all required origins:
Webhook Delivery Failures
Symptoms: Webhooks not received, timeout errors, SSL certificate errorsCommon Causes
| Error | Cause | Fix |
|---|---|---|
| Connection timeout | Endpoint too slow | Respond immediately, process async |
| Connection refused | URL not accessible | Use public URL, not localhost |
| SSL certificate error | Self-signed or expired cert | Use Let’s Encrypt or valid CA |
| DNS resolution failed | Domain doesn’t resolve | Check DNS configuration |
Timeout Requirements
| Webhook Type | Default Timeout | Max Configurable |
|---|---|---|
| StartWebHookPayload | 10 seconds | 300 seconds |
| ToolWebHookPayload | 5 seconds | 300 seconds |
| Other webhooks | 30 seconds | — |
Fix Slow Endpoints
Fix Slow Endpoints
Test Webhook
Test Webhook
Local Development
Local Development
Use ngrok to expose local endpoints:
Concurrent Call Limit Exceeded
Symptoms: “Concurrent call limit exceeded”, HTTP 429 with “Too many active calls”Limits by Plan
| Plan | Concurrent Lines | Minutes/Month |
|---|---|---|
| Developer (Free) | 1 | 1,000 |
| Growth | 30 (at start) | Contact sales |
| Enterprise | Custom | Custom |
Growth and Enterprise plan limits can be adjusted based on your needs. Contact support@dasha.ai to discuss your requirements.
Check Current Usage
Check Current Usage
Solutions
- Wait for slots — Poll active calls, schedule when capacity available
- Stagger scheduling — Space calls over time instead of bulk scheduling
- Upgrade tier — Contact support@dasha.ai for higher limits
Tool Calling Failures
Symptoms: Agent says “I’m unable to complete that request”, tool webhook errors, missing parametersCommon Causes
| Error | Cause | Fix |
|---|---|---|
| Timeout | Webhook takes over 10 seconds | Optimize endpoint, respond faster |
| Invalid response | Response not JSON | Return valid JSON object |
| Missing parameters | Schema mismatch | Check tool schema in agent config |
| Agent doesn’t use tool | Prompt unclear | Add explicit tool usage instructions |
Verify Tool Configuration
Verify Tool Configuration
Fix Prompt for Tool Usage
Fix Prompt for Tool Usage
Add explicit instructions to system prompt:
Voice Quality Issues
Symptoms: Robotic voice, choppy audio, audio dropoutsQuick Fixes
| Issue | Try This |
|---|---|
| Robotic voice | Switch TTS provider or try different voice |
| Too fast/slow | Adjust speed (0.8-1.2 range) |
| Choppy audio | Check network quality, use wired connection |
| Inconsistent quality | Try different voice ID |
Provider Comparison
| Provider | Quality | Best For |
|---|---|---|
| Cartesia | Good | Emotions, wide speed range |
| ElevenLabs | Excellent | Premium quality, voice cloning |
| Inworld | Very Good | Character voices, gaming |
| LMNT | Good | Consistent output |
Switch Provider
Switch Provider
ASR Transcription Errors
Symptoms: Agent responds to wrong words, user repeats themselves, numbers transcribed incorrectlyQuick Fixes
| Issue | Fix |
|---|---|
| Wrong language | Set correct primaryLanguage (e.g., en-US) |
| Background noise | Enable ambientNoise: true in features |
| Frequent mishearing | Add confirmation prompts to system prompt |
Enable Noise Suppression
Enable Noise Suppression
Add Confirmation to Prompt
Add Confirmation to Prompt
LLM Response Issues
Symptoms: Slow responses (over 5 seconds), irrelevant answers, hallucinations, repetitionQuick Fixes
| Issue | Fix |
|---|---|
| Too slow | Use faster model (Reflex-1, GPT-4.1 Mini) |
| Irrelevant responses | Improve system prompt clarity |
| Hallucinations | Lower temperature (0.3-0.5) |
| Repetitive | Adjust presence/frequency penalties |
Model Comparison
| Model | Speed | Intelligence | Cost |
|---|---|---|---|
| Reflex-1 | Very Fast | Good | High |
| GPT-4.1 Mini | Fast | Good | Low |
| Claude 3.5 Sonnet | Fast | Excellent | Medium |
| Gemini 2.5 Flash | Very Fast | Good | Low |
| GPT-4 | Medium | Excellent | Medium |
Adjust Settings
Adjust Settings
Call Recording Not Available
Symptoms: “Recording not found”, null recordingId, some calls have recordings but others don’tQuick Checks
| Issue | Fix |
|---|---|
| Recording disabled | Enable recordingEnabled: true in agent features |
| Just completed | Wait 1-2 minutes for processing |
| 404 on download | Use correct endpoint: /calls/media/{recordingId} |
Enable Recording
Enable Recording
Check Recording Status
Check Recording Status
Dashboard Loading Issues
Symptoms: Blank screen, infinite spinner, 403/401 errorsQuick Fixes
| Issue | Fix |
|---|---|
| Blank screen | Clear cache and cookies, try incognito |
| 401/403 errors | Log out, clear cookies, log back in |
| Works in incognito | Disable browser extensions |
| Still broken | Check status.dasha.ai |
When to Contact Support
Email support@dasha.ai when:- Service unavailable for over 15 minutes
- Data loss or security concerns
- Tried all troubleshooting steps without resolution
- Need account-level changes (tier upgrades, custom limits)
- Agent ID and Call ID (if applicable)
- Error messages (exact text)
- Steps already tried
- Timestamp when issue occurred
Related Resources
- Dashboard Testing — Test agents in the browser
- Call Inspector — Deep call analysis
- Webhooks Overview — Webhook configuration and security
- Production Checklist — Pre-launch validation