[ AI RECEPTION LAYER ]
No request
goes unanswered
A written and spoken reception layer for the enquiries that arrive out of hours, arrive at the same moment, or simply join a queue.
[ TRY IT ]
Reception, live
Ask a question, request an appointment, or try something the system is not meant to answer. The system panel shows what it did at each step and which record it produced.
Harbour & Vale
Sample configuration — professional services firm
This demo runs in your browser. The decision layer is rule-based; there is no language model behind this page. Nothing you type leaves it: no email, SMS, calendar or CRM action is taken, and no appointment is created. Both businesses are fictional sample configurations.
Voice mode is off on this page. Turning it on requires the server-side speech endpoints and a provider key.
What it does
- 01
Voice and written reception
A reception layer that knows your business, on the phone, the website and WhatsApp.
- 02
Capture and qualification
It answers questions, takes the contact details and qualifies the request by priority.
- 03
Calendar and booking
It reads availability and writes the appointment straight to the calendar.
- 04
A defined handover
Anything needing judgement, or outside its scope, goes to a person.
[ HOW IT WORKS ]
What a reception layer is built on
A reception layer is not a product on its own; it sits on top of an operation that already exists. Four things have to be settled before it is built.
- 01
The business knowledge base
Everything the agent may say is written down first: services, hours, how each question is answered, what is out of scope. This is not a text file — it is the boundary of the system. On anything unwritten, the agent does not speak.
- 02
Channels
Phone, website, WhatsApp and social messages share the same reception logic. Which channels you open depends on where the business actually receives requests; opening all of them is unnecessary for most.
- 03
Calendar and records
The booking flow depends on what the calendar it writes to returns. With the integration in place the system reads availability and creates the appointment. Without it, what it produces is an appointment request — not a confirmed booking. Those are different things, and the interface says so.
- 04
The handover rule
When a conversation goes to a person is defined up front: anything needing judgement, anything out of scope, anything asking for a pricing commitment, and anything the agent does not know. The handover is part of the design, not a failure state.
The difference between the demo and a deployment
The demo above produces an appointment request. In a real deployment the same flow connects to your calendar and the appointment is confirmed there — but until that connection exists the system keeps producing requests and creates no appointments. Not confusing the two is the most important distinction on this page.
- This demo
- Runs the conversation, structures the request and produces the record. It writes to no calendar.
- A real deployment
- The same flow connects to your calendar; availability is read and the appointment is confirmed there.
[ WHERE IT HELPS ]
Where we build it
A reception layer earns its place where requests are frequent and repetitive, and where every missed call has a cost.
- 01
Clinics and healthcare
Appointment requests, service questions and routing. The agent performs no medical assessment, makes no diagnosis and recommends no treatment; it structures the patient's request and hands it to the practice's own staff.
- 02
Estate agents and property
Listing questions, viewing requests and routing to the right negotiator. Qualification is the step that saves the most time here.
- 03
Service businesses
Out-of-hours requests captured, qualified, and waiting as a follow-up list in the morning. For most businesses this is the setup that pays back fastest.
[ BOUNDARIES ]
What it does not do
Regulatory compliance depends on the scope of a given deployment and on the business's own processes, and is agreed per engagement. This page is not a compliance undertaking.
- 01
It makes no medical diagnosis, recommends no treatment and takes no clinical decision.
- 02
It makes no pricing commitment and quotes no figure outside a written price list.
- 03
Without the integration in place it does not confirm an appointment — it produces an appointment request.
- 04
On anything it does not know it invents nothing; it hands the conversation to a person.
- 05
The demo on this page is rule-based and runs in your browser: no language model, and nothing you type leaves the page.