Case Study · Yubi
Four separate risk products. Four different interfaces. Bank teams toggling between all of them to assess a single borrower. I redesigned them into one unified decision engine — and validated the concept with seven banks before any production code was written.
Accumn is an AI-powered platform designed to help underwriters and risk management teams at banks and fintechs monitor borrower health and prevent fraud after a loan is disbursed. It's a significant product operating in a ₹30,520cr market.
The problem: it wasn't really one platform. It was four separate products — Explore, Insight, Fulfilment, and Risq — each built independently, each looking and behaving differently, and each requiring bank employees to sign in separately to get different data on the same borrower.
Four products, four separate UIs — a bank analyst assessing a single borrower had to context-switch between all of them.
With no formal user research channel, I had to find an unconventional path to the people who actually used the product. The Sales team had the only direct access to clients — so I started there.
Rather than asking for favors, I offered value first. I noticed their presentation decks were outdated and weren't doing the products justice. I redesigned them. The gesture built trust, and that trust opened doors: I was invited to join client calls, and eventually an in-person visit with a major bank.
Redesigning the Sales team's presentations wasn't the brief — it was the move that earned access to real clients.
With that framing locked, I designed a unified platform from the ground up. The goal wasn't to merge four products into one for the sake of tidiness — it was to turn a data dump into a decision engine. Three guiding principles shaped every screen.
The new dashboard leads with an Overall Risk Score — the most critical answer to the most critical question a bank employee has: "How risky is my portfolio right now?" Color and size communicate health at a glance, before the user has read a single number.
I replaced dense spreadsheets with an interactive treemap that lets analysts explore their loan portfolio by industry, geography, or risk tier — and drill into problems with a single click. The chart does the synthesis that used to happen in the analyst's head.
Each borrower profile now tells a story. The most critical signals — risk score, red flags, recent changes — lead. Supporting detail is available on demand. Users move through a borrower's profile in the order that matters for a decision, not the order the data arrived.
Once the prototype was ready, it became our most powerful sales tool. The reaction from potential clients was immediate.
This project reinforced something I'd been learning: the designer's leverage isn't just in what they design — it's in who they talk to and how they earn those conversations.