Skip to content

How it works

Put the team in the thread. Let the routine answer itself. Close it in the chat.

Every screen below is a working picture of the product. Scroll, click, press play. Nothing here describes a feature we could not show.

Chapter 1

The inbox

Every conversation, visible to everyone who should see it, assigned to one person at a time.

Live inbox · Saanjh · 2 numbers

A shared WhatsApp inbox: a customer asks about stock, an automation answers with the live stock in two seconds, the customer asks the price, and a suggested reply waits for a person, who takes over and sends it.
AAAarti+91 98••• •••42unassigned

Waiting for a message...

  • Assign, do not forward

    A thread belongs to one person at a time, with the history attached. Handover is a click, not a screenshot.

  • Several numbers, one inbox

    Sales and support on different lines, one place to read them.

  • Drafts and quick replies

    Half written replies are kept. The sentences you type ten times a day are one keystroke.

Chapter 2

Contacts and broadcasts

Segments built from tags in seconds, and a broadcast whose delivery, read and reply numbers come back while it is still sending.

A segment being built from tags; the audience count changes as tags are added.
Everyone who is

0 contacts

Saved as “Dhanteras · bought before”

A broadcast in progress: delivered, read and replied counters ticking up as the send goes out.
Template: dhanteras_meera_v2scheduled
Delivered
0
Read
0
Replied
0

Broadcasts use Meta approved templates and go out at a safe pace, so the number is never flagged for cutting corners. Contacts who have not opted in are skipped, and the record of who agreed is one you can produce.

Chapter 3

Automations

Draw the flow. Press play. Watch a pretend customer travel it.

When does the message arrive
An automation canvas with seven blocks: a keyword trigger, a business hours check, a template send, a wait for reply, an AI agent turn, and an assignment. Pressing play lights the path a customer takes.

Seven of about twenty block kinds. Press play to send a customer through.

The customer's side of the automation as it runs.
KAKabir

Waiting...

The blocks you draw with

  • Triggers

    a keyword, a new conversation, or a message from your other software

  • Send

    a text, an approved template, or an in chat form

  • Wait

    for a reply, or for a set time

  • Decide

    branch on what they said, switch on a field, check business hours

  • Reach out

    send the conversation's details to your other software

  • Ask Aura

    an AI agent turn, a product recommendation, an image, or approved campaign creative

  • Tidy up

    update the contact, assign a person, mark it resolved

Automations can also be imported from a file, so one that works for one number can be reused on another. If a block that asks Aura for something cannot get it, the automation skips that block and finishes the conversation rather than leaving a customer waiting.

Chapter 4

AI agents

Choose what the agent may know, and whether its replies wait for a person. Then test it before a customer ever sees it.

The agent editor: a preset picker, five grounding switches, tone and length, and a suggest or auto send toggle.
Start from
Preset
What it may know

from Aura · four fields

from Aura · first 40 products

from Aura · requested topics only

from Hella · this customer only

hours, delivery areas, policies

Tone
warm
Length
short

Every reply waits in the inbox as a draft until a person sends it. The default.

Handoff words: refund, complaint, manager, legal, angry. Always a person first.
A test conversation: the customer's question and the agent's reply, which changes as switches are toggled, with the sources it used.
SNTest against a real thread
Is the Meera set available, and can you deliver to Udaipur?

Suggested reply · waiting for a person

Yes, the Meera set is in the shop now. 18k, ₹1.2L to ₹1.6L depending on the stones. It comes hallmarked, with the certificate. Let me check delivery for Udaipur and come back to you. Want me to hold a fitting slot?

Brand voiceCatalogueIndustry guidance

Sources are shown on every reply. An agent that answers oddly is almost always one given the wrong context.

  • Presets you can edit freely

    Sales, support and catalogue Q&A are starting points. Once created, an agent is yours to change in every detail.

  • Grounding you switch on

    Brand voice, catalogue, industry guidance, order history, plus local notes for things like opening hours. Each costs a little per reply, so nothing is on by default that you did not choose.

  • Suggest before send

    Every preset starts in suggest mode: the reply waits in the inbox as a draft. Auto send is a decision you make per agent.

  • A human when it matters

    Refund, complaint, manager and words you add yourself go straight to a person, before the AI sees the message.

Chapter 5

Forms inside the chat

Bookings and sign ups as native WhatsApp screens, not a link people do not open.

A WhatsApp Flow: a native in chat booking form with three screens.
SJSaanjh · booking

Book a fitting

Rendered by WhatsApp itself. No link, no browser.

Chapter 6

Selling in the chat

Catalogue, order, payment, status. One thread from the first question to delivered.

  1. 01

    Sync your catalogue

    Products live in Hella and sync to Meta's product catalogue, so they can be browsed natively inside WhatsApp rather than sent as images.

  2. 02

    They browse and ask

    Customers open the catalogue in the chat, ask about an item, and get an answer from a person, or from an automation that knows the real stock and price.

  3. 03

    The order becomes a record

    What they choose becomes an order with a status your team moves through: placed, confirmed, shipped, delivered. Not a note in someone's head.

  4. 04

    They pay by UPI or card

    Payment is collected through Razorpay and tied to the order and the conversation, so reconciling what was paid for what is not detective work.

Your business profile, hours and address are managed from the same workspace, and a scannable code that opens a chat can go on packaging, print and the shop counter.

See it against your actual workflow

Tell us how you run WhatsApp today, one number or five, two people or twenty, and we will show you exactly what changes, including the parts where Hella would not help.

Chat with us