DCB CUBE 2.0 AI Platform
AI-powered onboarding for one of Indian banking’s most complex account types — a conversational platform that guides any relationship manager through opening a Trust account by chat, not forms.
The context
Opening a Trust account is one of the most complex onboarding journeys in Indian banking — layered across multiple Trust Acts, with dozens of document and KYC checks that change from case to case.
No specialist in the room
Relationship managers with no prior trust experience were expected to open these accounts correctly.
Volume at national scale
A single executive stakeholder oversees 600–700 trust-account openings a day, nationwide.
Manual at every step
Every account meant navigating regulation by hand, gathering the right documents, and verifying each one’s validity.
It didn’t scale
The result was a slow, error-prone, expert-dependent process — impossible to grow to national volume.
The brief: prove that an LLM-driven chatbot could guide any RM through the entire journey — accurately, and without a specialist in the loop.
Project overview
CUBE 2.0 is an AI-powered onboarding platform built for DCB Bank. At its centre is Ikee, a conversational assistant that lets any relationship manager open complex accounts through a simple chat — not forms — while the system handles the heavy lifting behind the scenes.
What CUBE 2.0 does
- Opens accounts through a guided conversation instead of dense forms.
- Walks any RM through each step in plain language — no specialist required.
- Reads and checks documents as they’re collected, flagging what’s missing or invalid.
- Produces the bank’s completed application paperwork at the end of the journey.
What I own
- Requirements — the single point of contact with DCB’s executive stakeholders.
- UX / UI — every screen and the full chatbot interaction model.
- Front-end — shipping the production UI, accelerated by code-gen automation.
- Scope — the platform behind DCB’s first wave of AI onboarding products.
The challenges
Four problems had to be solved at once for the concept to be credible to a bank.
Regulation overload
Multiple Trust Acts and case-by-case rules that no single RM could be expected to know.
Inexperienced operators
The product had to make a first-timer perform like a seasoned trust specialist.
Document validity
Every document had to be the correct one — and actually valid — before submission.
Built for scale
The journey had to hold up across roughly 70,000 customers nationwide.
Design goals
Make complexity invisible
Hide the regulatory machinery and surface only the next simple step the RM needs to take.
Guide, never gate
Ikee leads the way and explains as it goes — it never blocks the RM with jargon or dead ends.
Status you can trust at a glance
Always show where the account stands and whether each document is valid, pending, or rejected.
True to the DCB brand
Every screen built to the DCB brand book — navy, yellow, Arial — so it feels unmistakably like the bank.
Paper-free by default
The entire journey runs end to end digitally, with zero physical forms.
One platform, many account types
CUBE 2.0 is built so the same conversational experience can open very different account types. The Trust journey proved the model; the platform is now being extended across DCB’s onboarding suite — each new type plugging into the same intelligent core.
Products I’m leading
The first three account types under the signed BRD — each one mine end to end across requirements, design and front-end.
Individual Staff Account
Fast, guided onboarding for the bank’s own staff members.
Individual Savings Account
Resident individual savings onboarding, designed for everyday scale.
Proprietorship Account
Non-individual onboarding tailored to sole proprietors.
The logic engine
Before a single screen was designed, I mapped the entire trust-account decision logic — every branch, document gate and validation path the assistant would follow. This map became the framework the LLM engine and the chatbot UI were both built on.
Architecture I helped shape
Owning the requirements end to end earned me a seat at the architecture table. After the BRD was signed, I helped draft the platform’s logical architecture — the layered system that carries a conversation from the RM’s screen all the way to the bank’s core.
Visual design system
Every surface is built to the DCB Bank brand book, so the product feels unmistakably like the bank from the very first screen.
Bold for headlines and section titles, where hierarchy matters most.
Regular for body and narrow for fine print — used consistently across every screen.
A system, not a set of screens
Three rules keep every account type feeling like one product.
From design to code
I don’t just design these products — I ship them. Using Figma Make I generate front-end code straight from my designs, which is how I act as both lead product designer and principal front-end developer on this engagement.
Outcome & learnings
POC → signed client
A two-and-a-half-month proof of concept convinced DCB’s CXO team — a PO was signed and DCB officially became our client.
From POC to a product suite
The first BRDs followed, kicking off DCB’s first wave of official AI onboarding products.
Owning requirements opened doors
End-to-end ownership of the POC earned me the architecture seat on the official build, not just the design one.
Designed for real scale
Rolling out through September, the platform is set to ease onboarding for roughly 70,000 customers across India.