A receptionist that never misses a call.
Real-time voice AI that answers, qualifies, and books appointments — while the front desk watches it happen live.
Live demos are password-gated — request access via the contact page and the password comes back in minutes.
Small clinics and home-services firms lose bookings every time a call rings out — and every missed call is revenue walking to the next listing.
A two-sided voice assistant: a click-to-talk widget on the business site for callers, and a live console where staff watch qualification fields fill, the calendar update, and the lead land in the pipeline.
- A complete call, hello to booked slot, with two-way audio
- Qualification fields auto-filling while the caller speaks
- The calendar and CRM updating the moment the slot is confirmed
The challenge
For a dental clinic or an HVAC company, the phone is the sales channel — and it goes unanswered at lunch, after hours, and whenever the desk is busy. Voicemail converts almost nobody.
The hard part isn’t answering: it’s answering usefully. The assistant has to hold a real conversation, capture structured lead data from unstructured speech, and close the loop with an actual booking — not a callback promise.
The solution — three decisions
Two surfaces, one call
The caller gets a friendly widget on the clinic’s own website; staff get a console. Both watch the same conversation — that cross-surface handoff is the product.
Liveness as the design language
Everything happening right now — audio meter, state lamp, streaming transcript — carries one signal color — and nothing else moves. On video, the call feels unmistakably alive.
Booking is the finish line
The assistant offers real open slots mid-conversation and confirms by voice; the slot flips to booked on the staff calendar in the same second.
How it works
The demo implements this shape end to end with a simulated service layer — the “extend for production” section lists what swaps in for live deployment.
Product tour
Qualification fills itself as the caller talksAssistant state — listening, thinking, speakingWord-by-word live transcript
What the demo shows
- Two-way audio: the assistant genuinely speaks its answers aloud
- A scripted caller journey that is fully repeatable on camera
- The “live line” following staff to any screen during an active call
- A 14-day dashboard where every number reconciles with the call log
Under the hood
- Call engine drives a scripted state machine: listening → thinking → speaking, with word-level transcript streaming and speech synthesis
- Shared in-memory store with subscriptions — a booking mutates calendar, pipeline, and history in real time across routes
- Deterministic seeded mock data, generated relative to “today”: believable names, lumpy call volume, no future timestamps
- Per-app design system enforced from a single DESIGN-SYSTEM.md; light console with an ink-dark sidebar
Demo Dental Clinic is fictional and labeled as a demo on every screen. Nothing here is presented as client work: no client names, no outcome metrics, no testimonials. The proof is the running product — open the live demo above and check every claim.
What we’d extend for production
- Swap the scripted engine for a live ASR/LLM/TTS loop over WebRTC
- Real calendar backends (Google/Outlook) with conflict handling
- Barge-in handling and interruption-aware turn taking
Build something like this.
Fixed scope in writing before any money moves, demos during the build, and full code ownership at handover.