What languages are supported?
Approximately 30 languages including Afrikaans, Arabic, Armenian, Azerbaijani, Belarusian, Bosnian, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Marathi, Maori, Nepali, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tagalog, Tamil, Thai, Turkish, Ukrainian, Urdu, Vietnamese, Welsh.
What control do users have over AI responses?
- Knowledge Base: Upload documents or add website URLs to ground the AI in your content.
- FAQ Management: Add Q&A pairs for consistent answers to common queries.
- Deflection Settings: Define topics the AI should avoid.
- Welcome Flow: Customise the greeting and starter questions.
- Persona Prompt: Edit the underlying prompt that defines personality and behaviour.
What is the agent loading sequence?
When a visitor opens your agent, they see a clear sequence of labelled phases: Initial Prompt → Mic Permission → Setting Up → Almost Ready → Long-Wait Recovery. This replaces the previous single "Starting" message.
Can the agent take actions?
Yes. Your agent can answer questions and — for commerce-enabled Spaces — manage shopping carts and hand off to checkout. Shopify and WooCommerce are both supported. Via the Omniagent API, you can attach custom tools (functions) to trigger any backend workflow.
Lead Capture
Spaces now includes a Lead Capture feature that lets you collect visitor contact details inline during a conversation. When a visitor shows intent to be contacted or to escalate to a person, the agent triggers a short form directly in the experience. The form includes a request summary (pre-filled from the conversation and editable by the visitor), the visitor's name, and their preferred contact method (email or phone). On submission, the request is emailed to the address you configure. Lead Capture is off by default — toggle it on from the experience setup. See Integration & Embedding for setup instructions.
Advanced session configuration (Omniagent API)
-
Web search: Set
useWebSearch: trueto let the agent search the web during conversations for real-time information beyond its knowledge base. Enabled by default — setfalseto restrict the agent to its provided knowledge only. Website search accuracy and content relevance were improved in July 2026; the agent now briefly acknowledges when it is looking something up during longer searches. -
Idle timeout: By default, idle sessions close automatically after a timeout. Set
disableIdleTimeout: trueon the agent to keep sessions open indefinitely — useful for kiosk deployments or long-running integrations. - User context (externalClientProfile): Pass structured metadata about the end user per session — name, plan, company, etc. — so the agent can personalise its responses.
- Opening instructions (initialSpeech): Instruct the agent on how to start each session before the user speaks.
-
Session cost: Each closed session includes a
costfield in USD — the billed cost calculated from active minutes and your model configuration.
What integration options are available?
Shopify (via Storefront MCP), WooCommerce, WebRTC Web SDK (React, Vue, Angular, vanilla JS), WebSocket, VoIP (Twilio, Telnyx, Vonage), SIP, Kiosk (Napster Station), and iFrame embedding for Spaces.
What are the technical requirements?
- WebRTC for browser-based audio and video sessions.
- Standard web browser — no specialised software needed.
- Mobile compatible — works on standard mobile browsers.
- ~300ms response latency.
Is my data secure?
Yes. Napster runs entirely on Microsoft Azure and is SOC 2 Type II, GDPR, and CCPA compliant. Data is encrypted at rest and in transit (TLS 1.2+). Your data is never used to train models.