sensAI
First AI multi-agent trading intelligence platform in the Indian trading space.
The Problem
Trading platforms give you everything — and that's exactly the problem. Too much data, too fast, for users who just want one clear answer: should I act on this or not?
The brief that landed was simple on the surface: add an AI feature to help users make better trading decisions. What it didn't say was which decisions, for which users, through which interface, or what better actually meant.
That ambiguity was the first thing to solve.
My Role
End-to-end product design — from problem framing to shipped interface. Worked directly with the AI/ML team throughout, which meant the design decisions had to account for what the models could actually deliver — not just what looked good in Figma.
This wasn't a handoff project. It was a collaboration.
"Two agents. Two jobs. One experience."
The solution wasn't one AI feature. It was two distinct agents solving two distinct problems — a stock screening agent that let users search in plain English instead of technical filters, and a financial chat agent that answered market questions in real time.
The design challenge wasn't building either agent. It was making sure the user never felt the complexity of both.
The Decision That Mattered
Multi-agent systems have a handoff problem. The moment a user crosses from one agent to another, the experience can fracture — different tones, different interfaces, different logic.
The solution was a single unified interface with contextual tools. One entry point. The agents switch underneath. The user never notices.
This required close collaboration with the AI/ML team — designing not just for the happy path, but for edge cases, failure states, and the moments where AI confidence was low. Those moments needed design solutions, not just engineering ones.
The Result
sensAI shipped as part of Shoonya — a platform used by real traders making real decisions. The design reduced information overload at the moment it matters most, and gave non-technical users access to intelligence that was previously locked behind complex filters.
What I Learned
Designing for AI isn't about making the AI look impressive. It's about making the user feel capable.
The best moment in this project wasn't when the interface looked good — it was when a non-technical user searched for stocks in plain English and got exactly what they needed. The AI disappeared. The decision remained.
That's the job.