Frequently Asked Questions
Find answers to the most common questions about SipPulse AI.
General
What is SipPulse AI?
SipPulse AI is an artificial intelligence platform for building conversational agents that can communicate via chat, voice calls, and messaging channels like WhatsApp. The platform provides text generation, speech-to-text, text-to-speech, and structured data extraction capabilities.
What can I build with SipPulse AI?
You can build various types of conversational AI solutions:
- Customer Support Agents: Handle inquiries, troubleshoot issues, answer FAQs
- Sales Agents: Qualify leads, schedule appointments, provide product information
- Voice Assistants: Automated phone systems for inbound and outbound calls
- Chat Widgets: Embedded assistants for websites and applications
- WhatsApp Bots: Customer engagement through WhatsApp Business
Can I analyze existing call recordings?
Yes! SipPulse AI is not just for building agents - you can also analyze your existing call recordings:
- Speech-to-Text: Transcribe audio files with diarization (speaker separation)
- Structured Analysis: Extract insights, sentiment, topics, and custom data from transcripts
- Batch Processing: Upload multiple files for automated analysis
This is ideal for call center analytics, quality assurance, and extracting insights from historical conversations.
See Speech-to-Text and Structured Analysis for details.
Which languages are supported?
SipPulse AI supports multiple languages for both text and voice:
- The platform interface is available in English and Portuguese
- LLM models support 100+ languages
- Text-to-speech voices are available in 20+ languages
- Speech-to-text supports 50+ languages
The exact language support varies by model and provider.
Pricing & Credits
How does pricing work?
SipPulse AI uses a credit-based system where you pay for what you use.
Agent Pricing (most common):
| Agent Type | Platform Fee | + Underlying Services |
|---|---|---|
| Chat | Per output token | + LLM tokens (input/output) |
| Voice | Per minute | + LLM tokens + TTS characters |
The platform fee and underlying service costs (LLM, TTS) are charged separately and appear in different cards on the Dashboard.
Direct API Services (without agents):
- LLM tokens: Charged per input/output token
- Speech-to-Text: Charged per minute of audio
- Text-to-Speech: Charged per character generated
- Structured Analysis: Charged per execution based on input size
View current pricing at sippulse.ai/pricing. Pricing may vary by organization based on your plan.
How do I add credits?
- Click on your credit balance in the top-right corner of the screen
- Select Add Credits from the dropdown menu
- Enter the desired amount and click Go to Checkout
- Complete payment on the checkout page (credit card, Google Pay, etc.)
- Credits are added immediately after successful payment
Do credits expire?
Standard credits do not expire. Promotional credits may have expiration dates - check the terms of any promotional offers.
Can I get a refund?
Unused credits can generally be refunded within 30 days of purchase. Contact support@sippulse.com for refund requests.
Agents
What's the difference between Chat and Voice agents?
Both use the same underlying AI capabilities, but:
- Chat agents: Communicate via text (API, Chat Widget, WhatsApp)
- Voice agents: Can make and receive phone calls using SIP integration
- Hybrid: The same agent can often handle both, with voice settings enabled
How do I test my agent before deploying?
Use the Playground feature:
- Open your agent
- Select Chat Playground or Voice Playground
- Test conversations without affecting production
Why is my agent responding incorrectly?
Common causes and solutions:
- Unclear instructions: Be specific about expected behavior
- Wrong model: Some tasks require more capable models
- Missing context: Add relevant knowledge bases or tools
- Conflicting instructions: Review for contradictions
See Prompting Agents for best practices.
Can agents access external data?
Yes, through several methods:
- Knowledge Base: Upload documents for RAG-based retrieval
- API Tools: Configure HTTP endpoints the agent can call
- MCP Servers: Connect to Model Context Protocol servers
Technical
What LLM models are available?
SipPulse AI integrates with multiple providers:
- OpenAI (GPT-4, GPT-4o, GPT-4 mini, etc.)
- Anthropic (Claude models)
- Google (Gemini models)
- And more
Models are shown in the UI when configuring agents. Availability may vary.
What voice providers are available?
Supported text-to-speech providers include:
- OpenAI TTS
- Azure Speech
- ElevenLabs
- SipPulse (pulse-tts)
Each provider offers different voices and language options.
Is there an API?
Yes, SipPulse AI provides a comprehensive REST API:
You can also generate API keys in the API Keys section of the platform.
What are webhooks used for?
Webhooks notify your systems when events occur:
- Conversation started/ended
- Message received
- Agent invoked a tool
- Analysis completed
See Webhooks for configuration details.
Security & Privacy
Is my data secure?
Yes, SipPulse AI implements enterprise-grade security:
- Data encrypted in transit (TLS 1.2+) and at rest
- SOC 2 compliant infrastructure
- Regular security audits
- Optional data retention policies
Where is data stored?
Data is primarily stored in secure cloud infrastructure. Enterprise customers can discuss data residency requirements.
Can I delete my data?
Yes, you can:
- Delete individual conversations and threads
- Remove agents and their associated data
- Request full account deletion via support
Do you use my data for training?
No, customer data is not used to train AI models. Your conversations and data remain private.
Integrations
How do I connect to my phone system?
SipPulse AI supports SIP integration:
- Configure SIP in your agent's Deploy settings
- Use the provided SIP URI or set up External Registration
- Configure your PBX to route calls to/from SipPulse
See SIP Integration for detailed instructions.
Can I use WhatsApp?
Yes, through WhatsApp Business API integration:
- You need a verified Meta Business Account and WhatsApp Business Account (WABA)
- In the agent's Deploy settings, click Connect on the WhatsApp card
- Complete the Meta authorization flow (Embedded Signup) - no token management required
- If needed, register your phone number with a 6-digit PIN
The integration is handled automatically through OAuth. You don't need to manage tokens or API credentials.
See WhatsApp Integration for detailed prerequisites and setup steps.
How do I embed an agent on my website?
Use the Chat Widget:
- Go to your agent's Deploy settings
- Configure the Chat Widget
- Copy the provided JavaScript snippet
- Add it to your website's HTML
See Chat Widget for customization options.
Troubleshooting
My agent is slow to respond
Possible causes:
- Model selection (larger models are slower)
- Long instructions or conversation history
- Complex tool calls
- Network latency
Try:
- Using a faster model variant (e.g., mini or flash versions)
- Optimizing instructions
- Reducing conversation context length
Voice calls have poor quality
Check:
- Network connectivity and bandwidth
- SIP trunk configuration
- Codec settings on your PBX
- Voice model selection
I'm getting rate limited
Rate limits protect service stability. If you're hitting limits:
- Contact support to discuss higher limits
- Implement request queuing in your application
- Optimize to reduce unnecessary API calls
Use Your Own API Keys
If you have a good rate limit quota with an LLM provider (OpenAI, Anthropic, etc.), you can integrate your own API key in Account → Integrations. Benefits:
- Rate limits are applied directly by the provider, not by SipPulse
- Usage is still tracked in your Dashboard for monitoring
- Cost attribution shows $0.00 (you pay the provider directly)
Getting More Help
Can't find your answer?
- Documentation: Browse the full documentation
- Support: Email support@sippulse.com
- API Reference: api.sippulse.ai/reference
See our Support page for all contact options.