Bijak · Agri-trade · B2B marketplace

Five years later, they were still running the systems I designed.

My roleLed product and architecture for the app rebuild, in partnership with Bijak's co-founder and CTO, Jitender Bedwal
DurationAbout 5.5 months (2019, at seed stage)
PlatformOffline-first Android app
CompanyBacked by Sequoia's Surge, RTP Global, Bertelsmann and Omidyar Network. Went on to raise about $35M.
Bijak home screen in Hindi and English
The business

Bijak is a B2B marketplace that helps traders, wholesalers and food processors buy and sell agricultural commodities, track their transactions and find reliable trading partners across India.

A busy mandi with sacks of produce
The challenge

Built for the mandi, not the metro.

When I started working with Bijak in 2019, it was a seed-stage company whose core business had taken off over WhatsApp and was gaining users quickly. Jitender Bedwal, Bijak's co-founder and CTO, had built an early app in Flutter, but it couldn't scale with the business. Its users worked in mandis and on the road with low-end phones, patchy 2G–4G networks and batteries that went hours without a charge. Many used their phones mostly for calls and WhatsApp. Bijak needed an Android app rebuilt from the ground up that worked offline first, which at the time was an uncommon approach.

DeviceLow-end phones
NetworkPatchy 2G–4G
BatteryHours without a charge
Phone useMostly calls and WhatsApp
What I did
01

Started with the users, not the screens.

Before designing anything, we studied how mandi trade actually works: who the users were, how they paid, how credit and trust worked between traders, and whether the app would force them to change how they did business. It shouldn't.

02

Made the edge case the norm.

Poor connectivity and low battery weren't treated as exceptions. The app was designed around them. It stored all data locally, could be launched in aeroplane mode and worked anywhere, syncing when a connection was available.

Step 1Data stored on the phone
Step 2Works in aeroplane mode
Step 3Syncs when there's a connection
03

Designed for first-time app users.

Users chose their language up front (Hindi or English). Every screen had a single input, and a custom numpad replaced typing wherever possible. Nothing on screen was meant to overwhelm someone unfamiliar with apps.

Hindi or English One input per screen Custom numpad
Bijak sign-up screens with custom numpad
04

Brought offline trade online, without changing how people traded.

Traders could log payments and attach whatever proof they already had: handwritten slips, photos of invoices, photos of lorries, pages from their khatas. Bijak could verify these later. Over time, this built a record of each trader's reputation that the business could use to assess credit.

Bijak trade records and payment screens
Proof traders already had
Handwritten slips Photos of invoices Photos of lorries Pages from their khatas
05

Built the engineering and product processes.

I set up how the team worked day to day. We shipped two releases every week and tested every build against the low-end devices our users actually had.

2releases every week
100%of builds tested on low-end devices
The result
5 yrs

In 2024, five years on, Jitender called to tell me Bijak was still running on the systems and development processes I had put in place.

Traders recorded transactions on the app more often and in higher volumes
Bijak's operations team needed far less manual field work to collect trade and working-capital data
Reconciling balances and pending payments became simpler and more accountable
Junk data from the previous app (bad phone numbers, wrong commodity and location details) largely disappeared
Bijak went on to raise about $35M across its Series A and Series B rounds

Building for users others overlook?

Start a conversation ↗