How it works

Watch it read an enquiry the way a person would

The exchange on the right is not a recording. It is the same arithmetic the real pricing engine performs — paise-accurate, tax in basis points — playing out as it would on your own number.

MP

Mehta Packaging

online

WhatsApp

What the books recorded

2,000 pcs @ ₹23.00
₹46,000.00
Delivery — Bhosari, Pune
₹850.00
GST @ 18%
₹8,433.00
Invoice total
₹55,283.00

Step one

It works out what it knows, and what it’s missing

“rate for 12x12x8 box, 2000 pcs” is not a form submission. Ordder pulls it apart into the things a quote actually needs, and keeps what it learns across the whole conversation — so an answer given in message four is never asked for again in message five.

Product

Matched against your catalogue, including the way buyers actually abbreviate things.

Quantity

With the unit — and it knows nos, kg, sqft, mtr, quintal and the rest.

Delivery

City, area or pincode, so the delivery charge is the right one.

Timeline

When they need it, which often decides whether you want the order at all.

Specifications

Size, grade, thickness, finish — whatever your catalogue says matters.

Then one question

The single most important thing still missing — not a list of six. Asking one at a time is the difference between a conversation and an interrogation.

Step two

The AI never invents a price

This is the part people are right to be suspicious about, so it is worth being exact. The language model reads the enquiry and works out which item is being asked for. It does not decide what that item costs.

Once the item is matched, the money is fixed arithmetic against your own settings: your rate, your GST slab, your delivery rule, your minimum order quantity. Money is held in whole paise, tax in basis points, so nothing drifts by a rounding error.

Where each number comes from

Which item
AI — matched against your catalogue
Rate per unit
Your catalogue
Quantity
The buyer
Minimum order qty
Your catalogue
Delivery charge
Your delivery rules + the buyer’s location
GST
Your tax settings
Total
Arithmetic

Step three

What stops it saying something stupid

It hands over when it is unsure

A negotiation, a product it cannot match, an unusual delivery — the deal goes to a named person with the whole thread attached, and they are told in the app and by email straight away. Not a silent queue somebody has to remember to check.

Approve-first, if you want the last word

Turn it on and the AI drafts but never sends. Nothing reaches a buyer until you have read it. Plenty of sellers run this way for the first fortnight and then switch it off.

It stops asking after three tries

If a conversation is not converging, it escalates rather than looping. A bot that asks a fourth question has already lost the buyer.

It only quotes what you published

Draft and archived catalogue items are invisible to it. If you have not marked something live, it will not be sold.

Nothing fails silently

A reply that cannot be delivered is retried and surfaced in the app with the reason. No buyer message is ever quietly dropped — we have been burned by that and built against it.

Getting started

An upload, not a project

  1. 01

    Upload your price list

    The Excel or CSV you already keep. Ordder works out which column is the rate and which is the GST slab, shows you exactly what it will save, and waits for you to confirm. A sheet with only a name and a rate is enough.

  2. 02

    Connect a channel

    Your WhatsApp business number, an email address, or a form snippet for your own site. Any one of them is enough to start; most sellers add the others within a week.

  3. 03

    Send it a real enquiry

    Forward something a buyer actually sent you and watch what comes back. That is a more useful test than any demo we could give you.

Your data

Your rates are not a shared resource

Isolated by the database

Every client’s rows are separated by PostgreSQL row-level security, and the app connects with a role that cannot bypass it. A bug in our code still cannot show your prices to another seller.

Hosted in India

Google Cloud, Mumbai region. Your catalogue, your buyers and your invoices stay in the country.

Credentials encrypted at rest

Your WhatsApp and Zoho tokens are AES-256-GCM ciphertext in the database, with the key held separately. A database dump on its own is worthless.

The full detail is in our privacy policy, including every sub-processor and what each one sees.

The honest test takes ten minutes

Upload your rate list, forward one real enquiry, read what it sends back. If it is wrong you will know immediately — and so will we.