A B2B ordering app that gives sales reps back their time to sell
The order jotted down in the store and typed up at night is now closed during the visit, with each store owner's offer and credit at hand.
In this case study
Executive summary
To take the order decision to the store and give sales reps back their time to sell, we built Reppos, the ordering app Reckitt and Nivea sales reps use with small retailers.
With a server of its own between the app and the e-commerce platform, the rep accesses each store's terms and an AI-enriched catalog, and builds orders in the stores without depending on a connection.
Today, 8 in 10 sales reps who place an order through the app come back to order through it again: the app has become part of the routine of people who sell in the field, and it already closes almost half of the orders with small retailers.
The challenge
At Reckitt and Nivea, small retailers are served by distributors' sales reps, field salespeople who visit up to 18 stores a day. With only a few minutes between one visit and the next, the rep needs to catch up on the store owner and the catalog and prepare the conversation: what to offer, with which promotion, under which rules and within what credit.
As is common in serving small retailers, this information did not reach the store with the rep, and the order went on paper and, typed up at night, could take up to 40 minutes per order. With an ordering app that brings to every visit all the information the rep needs, we gave them back their time to sell.
The case in numbers
- 8 in 10
- sales reps who place an order through the app in a month order through it again the next month
- More than 700
- business rules, such as minimum order, credit, discount budget and promotion, written down and linked to the code
- 36 seconds
- for a photo of a paper order to become a cart in the app
- +93%
- in orders through the app over the previous year
- Less than 1%
- of the value that goes through the app is the cost of supporting and evolving it
Numbers like these start in the field, understanding how each channel buys. Let’s talk
Inside the technology
The immersion in the sales rep's journey showed that the app could not depend on a connection, because it had to serve reps in remote places too, where the signal fails. So the journey in the app starts before the store: the rep chooses the day's visits, and the app downloads the catalog and the terms for each one.
A server of its own sits between the app and Reckitt's e-commerce platform. It prepares this catalog for offline mode and, when there is signal, fetches price, stock and order terms from the platform in real time.
On top of this base comes AI ordering. A photo of the paper order, or the store owner's PDF or spreadsheet, becomes a cart, and a photo takes 36 seconds. The decision stays with the rep, who checks every item before closing.
The solution
An app that speeds up the work of distributors' sales reps, from the visit to the closed order.
Store profile
Shows that store's promotions, credit, minimum order and price per payment method, before the rep opens the cart
What to offer
Brings suggested lists and repeats the last order
AI ordering
Turns a photo of the paper order, or the PDF or spreadsheet the store owner sent, into a cart in less than 1 minute; the rep reviews, swaps or deletes each item before closing
Substitute for out-of-stock items
Suggests, based on similarity between products, what to offer in place of the out-of-stock item
On-the-spot negotiation
Applies the rep's discount budget (Verba Flex) and the additional discount, within each one's limits
Payment
Offers Pix (Brazil's instant payment), boleto (bank slip) paid upfront and payment on delivery
Offline mode
Keeps each store's catalog and terms on the device, updated every 24 hours; the order is sent when the connection returns
The results
The rep decides the order in front of the store owner, and the night is no longer for typing. In 2025, orders through the app grew 93% over the previous year, and the value ordered, 113%. Starting from zero, the app already closes almost half of the orders with Reckitt and Nivea's small retailers.
Even with a new version every 2 weeks, supporting and evolving the app costs less than 1% of the value ordered through it.
Which order in your operation is still decided far from the buyer? In a few weeks, we design with you the app that takes the decision there. Let’s talk
How we worked
The work began in the field, interviewing sales reps and distributors about their day. That is where the app's design came from: to sell better in each store, the rep needed the right information at hand, even more than lower prices.
The first wave was about mobile usability, because neither the rep nor the store owner spends the day in front of a computer. Then came the intelligence about each store and, in the third wave, AI ordering.
In the most recent wave, we rebuilt the app and the server with AI agents, inside an environment with the project's rules (harness engineering): every change is planned first, goes through automated checks and is reviewed by an agent without the task's context. With these changes, the app gained 2.4 times more automated tests and more than 700 business rules written down and linked to the code.
Today the team delivers in about a third of the time it used to, with the front-end and back-end specialists looking after the harness and the skills and supervising the code of every delivery. In our projects, lean teams of experienced people, supported by AI agents, design the rules, write the code and speed up every delivery without giving up quality.
Questions this case answers
Question: A dedicated app for the sales force, or the e-commerce platform's website on a phone? Answer: For the sales rep who decides the order in the store, the dedicated app. The website serves the store owner who buys on their own; the rep needs each store's terms, suggestions of what to offer and the ability to capture the order even without signal. At Reckitt, the app went from zero to almost half of the orders with small retailers.
Question: What changes when the development of a B2B app becomes AI-assisted? Answer: Changing becomes safer and faster. At Reckitt, rebuilding the app and the server with AI agents left the app with 2.4 times more automated tests and more than 700 business rules written down and linked to the code, and new features reach the sales rep faster.