Your business, mirrored
Every conversation on WhatsApp, Instagram, email or marketplace updates a living twin of your customers, products, inventory and orders. Its agents act on it.
Priya Nair
2mNeed 4 blue XL polos before Friday
@arjun.styles
14mIs the linen shirt back in stock?
XYZ Uniforms Co.
1hPO #4471 · 120 units, net 30
Rahul M.
3hExchange size M → L, order 2291
Fabrico Mills
5hLead time now 11 days on cotton
Hi! Need 4 blue XL polos before Friday for our store opening.
I have 4 × Polo Classic in Cobalt, XL reserved for you. ₹1,196 each, ₹4,784 total. Dispatch today, delivered Thursday. Want the invoice?
Yes please. Do you have matching chinos?
Your customers keep using the channel they already use. Lipi becomes the operational brain behind it.
How it works
Conversations stop being chat logs and start being state
- 01
A message arrives
Any channel, normalised into one canonical shape by a thin adapter.
- channel
- from
- +91 98 ··· 4410
- text
- "4 blue XL polos by Friday"
- 02
Intent is extracted
Not filed as a chat log. Parsed into intent, entities, quantities and urgency.
- intent
- buy
- qty · size · colour
- 4 · XL · cobalt
- deadline
- 2026-08-28
- 03
Twins are updated
Customer, product, inventory and order twins upsert, and an immutable event is appended.
- customer_twin
- bulk_buyer = true
- inventory_twin
- reserved 4, remaining 0
- order_twin
- ord_2318 quoted
- 04
Agents act on it
Sales quotes, procurement drafts the restock, finance raises the invoice.
- sales
- quote sent, 1.2s
- procurement
- PO draft, needs approval
- finance
- INV-4482 issued
A living model, not a database of rows
Each twin evolves continuously and links to the others. That is what lets an agent answer “4 blue XL polos before Friday” with a real reservation instead of a plausible sentence.
Customer Twin
Buying history, size profile, negotiation style, price sensitivity, payment terms, return reasons and predicted next purchase.
Lifetime value
₹1,84,200
Orders
27
Return rate
4.1%
Product Twin
Variant-native, not a flat SKU list. Size by colour, with per-variant stock, margin and a cross-sell graph.
S · M · L · XL · XXL by cobalt / olive / white
Inventory Twin
Live stock across warehouses, reserved units, incoming shipments and forecast shortages.
- Mumbai74
- Surat41
- In transit18
Order Twin
One object from conversation to quote, payment, packing, delivery and returns.
- Quoted
- Paid
- Packed
- Shipped
- Delivered
Supplier and Knowledge Twins
Response time, lead time, defect rate and MOQ on one side. Policies, warranty terms and sizing charts on the other, so agents answer from your rules rather than from guesswork.
On-time
96%
Avg. lead
6d
Defect rate
0.8%
Policies indexed
34
Nine specialists, one source of truth
Agents do not pass messages to each other. They read and write the same twins, which is why their actions stay consistent and why every one of them is traceable back to an event.
Revenue
Sales
Quoting, negotiation, cross-sell
Marketing
Campaigns, abandoned-cart recovery
Analytics
Forecasting, KPIs, customer insight
Operations
Inventory
Availability, reservations, replenishment
Procurement
Supplier quotes, purchase orders
Logistics
Shipping, tracking, delivery updates
Trust
Support
Complaints, returns, warranty
Finance
Invoices, reminders, reconciliation
Compliance
Policy checks, audit trails, approvals
Nothing reaches a customer unapproved.
Every agent action lands in a queue until you trust it. Approve individually, approve a whole action type, or let an agent run unattended once it has earned it.
Why a twin
The difference is what the AI is standing on
Another omnichannel inbox
Lipi AI
Bring your channels. We'll build the twin.
We're onboarding a small number of apparel businesses first. Connect WhatsApp and your catalogue, and see your twin populate in an afternoon.