Conversational commerce automation
A WhatsApp shopping agent that recovers carts and answers sizing at 2am
A fashion brand selling across India and the Gulf did most of its customer conversation on WhatsApp, with a team of eight answering sizing, delivery and exchange questions by hand. Abandoned carts were followed up with a generic blast. We built a WhatsApp agent that knows the catalogue, the customer and the order, recovers carts with a conversation, and hands off to a person when it should.
- Client
- Fashion and footwear D2C brand
- Region
- India and Middle East
- Scale
- 70 to 110 employees
- Timeline
- 7 weeks
Representative engagement. Client details are anonymised and figures reflect typical outcomes from comparable deployments.
What the client got
0%
Abandoned cart recovery rate
Up from 6 percent with the generic discount blast.
0%
Of conversations resolved without an agent
Sizing, delivery, stock and exchanges handled in chat.
0%
Less discount given on recovered carts
Offers go only where the conversation shows price was the blocker.
The bottlenecks
Bottleneck 1
Sizing questions decided the sale
Customers asked “will a 38 fit me if I wear 7 in another brand” hundreds of times a day. Slow answers meant abandoned carts.
Bottleneck 2
Cart recovery was a blast
Every abandoned cart got the same discount message. Margin was given away to customers who would have bought anyway.
Bottleneck 3
Exchanges clogged the queue
Size exchanges were a multi-message negotiation that occupied agents and delayed answers to new shoppers.
What we built
Step 1
Catalogue-grounded agent
The agent answers from live product data, size charts, brand fit notes and return policy. It compares sizes across brands using the brand’s fit guide, suggests alternatives when stock is out, and sends product cards the customer can buy from.
Step 2
Conversational cart recovery
Abandoned carts trigger a message that asks what held the customer back. Sizing doubts get a sizing answer; delivery questions get a date; price objections get an offer only if the customer’s history justifies it.
Step 3
Self-service exchanges
The agent checks eligibility, confirms the replacement size is in stock, books the pickup and issues the new order, all in the chat.
Step 4
Human hand-off with context
Complaints, damaged goods and anything the agent is unsure about go to a person with the conversation summarised and the order open.
The automation workflow
WhatsApp commerce flow
From message or abandoned cart to order
Questions clients ask about this engagement
Does the agent speak Hindi and Arabic?
Yes. It responds in the language the customer writes in, including mixed-language messages common on WhatsApp.
Can it give discounts on its own?
Within rules the brand sets: maximum percentage, customer eligibility and category exclusions. Anything outside the rules goes to a person.
More in E-Commerce & D2C
Industry overview
AI email marketing automation
An AI email marketing engine that ships campaigns in hours, not weeks
An agency running email for a dozen consumer brands was limited by how fast its team could brief, write, build and QA campaigns. We built an AI engine that turns a product feed and a campaign brief into segmented, on-brand emails ready for review, and automates the lifecycle flows every brand needs.
5x
More campaigns per brand per month
Customer support automation
Order status, returns and exchanges handled without a support agent
Three quarters of support tickets were about order status, returns or exchanges, and the queue doubled every sale. We automated those tickets end to end, connecting the help desk, Shopify and the carriers, and left agents with the conversations that need a person.
72%
Tickets resolved with no agent
Catalogue and content automation
Product listings for 40,000 SKUs written, translated and kept compliant by AI
A seller with 40,000 SKUs on Amazon, eBay, Otto and its own store could not keep listings complete, consistent or translated. New products took weeks to go live. We built a content pipeline that turns supplier data into optimised, marketplace-compliant listings in six languages, with humans reviewing exceptions rather than writing from scratch.
2 days
From supplier feed to live in six languages
Wholesale and retail lead generation
Taking a D2C beverage brand into 300 retail doors with outbound
A direct-to-consumer beverage brand had strong online reviews and rising ad costs. Growth needed wholesale: independent grocers, gyms, cafés and regional chains. We built an outbound engine to buyers and owners, with samples and sell-sheets handled by workflow, and the brand opened hundreds of retail doors without a field sales team.
312
Retail doors opened
Have a similar bottleneck?
Tell us how the work happens today. We will show you the workflow we would build and prototype it on your own process.