How It Works

Your AI Receptionist in Minutes

No technical skills, no hardware, no app downloads. From setup to live in minutes — here is exactly how it works and what makes the platform reliable.

Setup

Live in three steps

Get your AI receptionist up and running without a single line of code or piece of hardware.

1

Tell us what you need

Choose your AI's name, voice, business hours, and what calls should handle — appointments, messages, or lead qualification.

2

Go live in minutes

We set up your dedicated AI receptionist on your own subdomain. Share the link on your website or send it directly to your customers — done.

3

Never miss a call

Your AI answers every call, captures everything, and sends you instant notifications. Review calls and transcription in your dashboard.

Technical Architecture

Built for reliability

An AI that answers your calls has to be dependable under real-world conditions — noisy lines, dropped connections, traffic spikes. Here is how the platform handles the edge cases most demos quietly ignore.

hearing
Low-Confidence Routing

When speech is unclear or noisy, the engine doesn't guess. It asks the caller to repeat once, then falls back to an on-screen secure form so details are captured cleanly instead of mis-transcribed.

cached
Stateful Context

Conversations survive dropped connections. A session-resumption handle plus sliding-window context compression lets a call reconnect and continue mid-sentence rather than restarting.

encrypted
POPIA-Native Architecture

Caller names, contact details and full transcripts are encrypted at rest with AES-256-GCM. The encryption key never touches the database. Privacy is built into the schema, not bolted on.

fingerprint
Idempotent Capture

Every captured submission is keyed to its live session token and linked to exactly one conversation. Retries and reconnects can never double-book or duplicate a lead.

shield
Circuit-Breaking & Failover

Upstream AI calls run behind a circuit breaker; email delivery retries across providers and routes. A single dependency hiccup degrades gracefully instead of dropping the call.

savings
Hard Cost Ceilings

Per-client and platform-wide spend gates are enforced inside a row-locked transaction before any minute is billed. The flat monthly price has a real, enforced cap — not a promise.