AI Voice Assistant and AI-Native Design for the Largest Indian Non-Bank Lender
Overview
Euristiq worked with a leading Indian non-bank lender to redesign its loan mobile app to an AI-native UI — pairing AI-native development and mobile app engineering to create a conversational, voice-first borrowing experience with no equivalent in the Indian lending market.
Our client, one of India’s largest non-bank lenders in secured loans,
wanted to use AI for the benefit of its client base. Rather than iterate on the existing app’s rigid, multi-step origination flow, Euristiq’s team followed a custom AI-native development Discover, Design, Deliver process to design and prototype a genuinely AI-native experience — one where the interface is generated by conversation rather than fixed screens — and shipped a working proof of concept in roughly six to eight weeks, within a three-month engagement.
A rigid, multi-step loan flow was starting to look like every other lending app on the market
Rigid, multi-step origination
Borrowers tapped through fixed screens in sequence — a structure that couldn’t flex around how people actually asked for a loan.
A generic, one-size-fits-all interface
Static forms and fixed screens gave every borrower the same interchangeable experience — nothing about it signaled category leadership or set the app apart from competitors.
Manual, in-person verification
Confirming collateral meant an agent physically visiting to photograph the applicant’s home or assets — slow, costly, and hard to scale.
A leader-in-AI native design
The client saw itself as a market leader and wanted to set the tone for generative AI in Indian lending — ruling out a conventional, incremental redesign and moving to the AI-native design.
Multimodal AI voice assistant interface
Key outcomes we delivered
A conversational, voice-first loan flow
AI-native design enables users to talk to the app instead of clicking through it. It’s powered by Google’s multimodal live model (the Gemini Live API, known during the project by its earlier name, “Google Live”) — handling real-time voice-in, voice-out conversation plus image-based interaction.
Generative UI that builds itself around the conversation
Instead of fixed pages, the interface generates a screen on the fly based on what the user says or asks — there’s no static page structure to navigate, and no menu to get lost in.
Memory-driven personalization across sessions
This assistant with AI-native UI remembers who the user is and their preferences across sessions — what the team calls a “GNI” (generative, personalized interaction) — so returning borrowers get tailored responses rather than a cold start every time.
A cross-sell layer built into the conversation
If a borrower asks for a loan to buy a phone, the system doesn’t just disburse funds — it surfaces relevant phone offers in the same flow, pulling the user deeper into the client’s ecosystem and lifting qualified-conversion opportunities.
Human-in-the-loop for legally binding steps
The AI voice assistant can recommend next steps — like suggesting a phone offer — but can’t autonomously execute anything legally binding, such as signing a contract. Those steps stay human-owned; the system works as a decision-support layer, not an autonomous agent.
From Progressive Web App to a native, AI-driven experience
React Native chosen over PWA for real-world Android performance
Native delivery integrated with the Gemini Live API for real-time voice
Discovery documentation that changed the client’s technical direction
A working proof of concept, not just a pitch deck
POC in ~6–8 weeks
A working proof of concept where users could speak to the app and receive live, spoken responses.
Voice, video and text
Multimodal and multilingual by design, responding in whatever language the borrower speaks.
Generative UI, not fixed screens
The interface is generated live from the conversation instead of a static page structure.
Cross-sell built into the flow
Relevant offers (e.g. phone deals) surface naturally inside the loan conversation.
Human-in-the-loop by design
AI recommends workflow steps; people handle anything legally binding.
AI-powered delivery
An AI Solutions Architect led the engagement — including a dedicated AI workshop for the client’s team — and the POC itself was built using AI-assisted development tools as part of an AI-powered SDLC.
Move beyond AI features.
Build AI-native
AI-native product experience, we can help you turn the idea into a working product.
FAQ
What is an AI-native interface?
In the AI-native interface, AI generates the experience itself via conversation, voice, or generative UI. So, it is different from the traditional app of fixed screens and menus. In this case, Euristiq built the lending flows around Google’s Gemini Live API so the interface responds to speech in real time instead of routing users through static forms.
Why use an AI voice assistant instead of a text-based chatbot for lending?
For a large user base, an AI voice assistant removes the requirement to read or type, making multimodal, voice-first interaction the more inclusive design choice. That was the core driver behind this project: replacing a form-based flow with speech, video and text support in whatever language or dialect the borrower speaks.
What does an AI-native development engagement with Euristiq look like?
An AI-native development engagement starts with AI strategy workshop where we pick the use cases relevant to your organization. Then we start scoping the target experience, mapping technical constraints (device hardware, connectivity, accessibility needs), and agreeing on what a working proof of concept needs to prove. From there, a senior team designs the conversational architecture and generative UI, and delivers a demoable PoC — typically within six to eight weeks — before scaling toward production. Share your scope, and we’ll come back with a recommended approach and timeline.
Recent cases
Modernizing Investment Portfolio Management Software for £12.9B in Assets
Digital Identity Verification with 100% PII Protection
