AI Systems That Run The Work
AGENRYX
All case studies
Product buildAI & Automation

WhatsApp AI Receptionist

An AI front desk that knows when to hand over

A multi-tenant WhatsApp receptionist designed to answer from approved business knowledge, understand voice notes, qualify leads and stop immediately when a human needs to take control.

Live system map
WhatsApp AI Receptionist

How the work moves

One connected path from request to result.

Client

WhatsApp AI Receptionist

Agenryx system

AI & Automation

Operating change

Text, voice notes, images and documents enter one normalised conversation flow

24/7

Inbound enquiry coverage target

4

Supported inbound content types

1

Durable handoff state the AI cannot override

01 · The challenge

What the client needed to resolve

The problem before Agenryx.

Small businesses lose enquiries outside office hours, but an unrestricted chatbot creates a different risk: invented answers, missed complaints and automation that keeps talking after a staff member should take over.

The change

From bottleneck to working system.

Before

Manual and fragmented

Small businesses lose enquiries outside office hours, but an unrestricted chatbot creates a different risk: invented answers, missed complaints and automation that keeps talking after a staff member should take over.

Agenryx built

One connected workflow

The v1 product design covers Meta onboarding, inbound processing, retrieval, lead classification, voice-note handling, human escalation, a conversation and knowledge dashboard, usage logging, security controls and production soak testing.

Resolved

A clearer way to operate

Text, voice notes, images and documents enter one normalised conversation flow

02 · What we built

The system behind the experience.

Architecture and workflow

Signed Meta webhooks feed a tenant-isolated message pipeline. Retrieval is scoped to the business knowledge base, message IDs enforce idempotency, voice notes are transcribed into the same flow, and handoff state hard-stops further automated replies.

What shipped or was designed

The v1 product design covers Meta onboarding, inbound processing, retrieval, lead classification, voice-note handling, human escalation, a conversation and knowledge dashboard, usage logging, security controls and production soak testing.

03 · Product walkthrough

See how the workflow moves.

Select a step or let the walkthrough advance automatically.

agenryx.system/whatsapp-ai-receptionist
System in action

The work enters one clear workflow

Small businesses lose enquiries outside office hours, but an unrestricted chatbot creates a different risk: invented answers, missed complaints and automation that keeps talking after a staff member should take over.

One source of truth
Clear ownership
Visible status

04 · What changed

The value is in the operating change.

Text, voice notes, images and documents enter one normalised conversation flow

Low-confidence, complaint, booking and payment cases can trigger a durable human handoff

Contacts, conversations, lead status, summaries and usage remain visible in one dashboard

Technology used

WhatsApp Cloud APIFastAPIPostgreSQLNext.jspgvectorOpenAI

Start with what you need

What would you like us to build or improve?

Next case study: Flavour Heaven