Frequently Asked Questions
Everything you need to know about setting up, billing, security, and deploying Speekee operators on your lines.
Our synthesis pipeline combines state-of-the-art text-to-speech engine layers. It naturally manages speaker breaths, pauses, filler words ("uhm," "so"), and accepts interruption (barge-in) as normal conversational behaviour. Most callers do not recognise they are speaking with an AI.
When the operator encounters a query it cannot resolve through its prompt parameters or action triggers, it will seamlessly invoke a SIP transfer to your human receptionist or support line, passing along the summarised transcript of the dialogue so far.
We support LM Studio (running locally on your server) and OpenAI APIs natively. Any custom LLM provider that exposes an OpenAI-compatible completion schema can be linked by configuring the endpoint URL and API key in the Voice Agents settings tab.
Yes. Audio streams and API payloads are encrypted in transit using SSL/TLS, and database records are encrypted at rest. Tenants can configure retention policies for voice traces and logs in compliance with local regulations. Audio buffers are purged immediately after call completion.
Yes. Speekee supports number porting and SIP trunk forwarding. You can forward calls from your existing business line to your Speekee webhook URL — no number change required for your customers.
Book a live demo using the button in our navigation bar. Our team will walk you through agent configuration, integrations, and launch in under 30 minutes. Most clients are live within the same business day.
Speekee supports configuring a secondary fallback LLM endpoint. If the primary endpoint times out or returns an error, the gateway automatically retries the fallback. As a last resort, the system gracefully transfers the call to your human agent queue.