🎯 Workshop Submission

Creative Use of AI Activity

Optional 20-Minute Activity — Team submission for Hisaab: WhatsApp Micro-Financing for Small Vendors

Goal
AI as multi-role collaborator
Outputs
Use Case · AI Feature · Tech Plan
Product Name
Hisaab (हिसाब)
Hisaab Activity Banner
1
Refined Use Case

Product Thinking: Who, What, and Why

Who is the user?

Small vendors such as street food sellers, fruit sellers, kiosk owners, and neighborhood shopkeepers who run cash-based businesses and already use WhatsApp daily.

What problem is actually being solved?

These vendors often need small amounts of urgent working capital to restock inventory, buy supplies, or manage short-term cash gaps. Traditional loans are too slow, paperwork-heavy, and often inaccessible because vendors may not have formal credit history or comfort with banking apps.

Refined Use Case

A small vendor needs a short-term loan of ₹5,000–₹20,000 to restock goods before the next business day. Instead of visiting a branch or downloading an app, they message a WhatsApp assistant, explain their need by text or voice note, share a bill photo or simple business details, and receive a fast eligibility decision and repayment plan.

2
AI Feature

What AI Actually Does in the Product

AI acts as a multi-role WhatsApp financing assistant that handles intake, understanding, decision support, and repayment guidance. Six distinct AI tools work in sequence:

🎙️
Speech-to-Text

Understands voice notes in local languages — no typing needed

📸
OCR Extraction

Reads bill photos, receipts, and transaction screenshots automatically

🌐
Multilingual NLP

Understands natural-language requests, intent, amounts, and urgency

📊
Cash-Flow Estimation

Estimates real working-capital need from business signals

🛡️
Risk Scoring

Builds Trust Score from alternative data, not formal credit bureaus

📅
Repayment Recommendation

Suggests plans calibrated to daily income, not monthly cycles

✦ Differentiated AI Feature: Alternative Trust Scoring

Instead of relying only on formal credit scores, the product builds a lightweight creditworthiness view using conversational consistency, uploaded business proof, prior repayment behavior, spending patterns, and cash-flow signals. This makes financing accessible to vendors who are usually invisible to traditional lenders.

3
Mini Technical Plan

Stack, Data Inputs, and How AI Fits

📱
Interface

WhatsApp Business API chatbot — onboarding, application, reminders, support

⚙️
Backend Stack

Loan workflow engine: KYC service, underwriting, disbursement, and collections system

📂
Data Inputs

Chat responses, voice notes, bill/receipt images, KYC details, repayment history, optional bank or wallet transaction data

🧠
AI Layer

STT, OCR/document extraction, multilingual NLP, cash-flow estimation, risk scoring, repayment recommendation engine

👤
Human-in-Loop

Low-confidence applications and fraud-risk cases routed to credit officer dashboard for manual review

🏦
Regulatory Layer

NBFC/bank partner holds lending license. Hisaab is the UX + AI layer. Compliance built in from day one.

4
One Minute Demo

Demo Script — Meet Ravi

Meet Ravi, a fruit vendor who needs ₹8,000 tonight to restock inventory for tomorrow morning. He opens WhatsApp and messages our loan assistant. In his own language, he sends a voice note saying what he needs, uploads a photo of today's supplier bill, and answers three simple questions. The AI understands the request, extracts details from the bill, estimates how much working capital he actually needs, and checks his Trust Profile using alternative business signals instead of a formal credit score. Within minutes, Ravi gets a recommended loan amount, a repayment schedule he can understand, and a next-step approval decision — all inside WhatsApp.

What the demo shows

  • Voice note → transcription → intent extraction (STT + NLP)
  • Bill photo → OCR → amount and vendor details extracted automatically
  • 3 follow-up questions asked by the AI (like a loan officer would)
  • Trust Score generated from alternative data — no CIBIL lookup
  • Recommended loan amount + repayment plan delivered in plain language
  • Decision delivered in under 2 minutes, no branch, no app
5
Judging Notes

How This Meets Every Slide Requirement

✅ Use multiple AI tools

Six distinct AI tools

STT · OCR · NLP · Cash-flow estimation · Risk scoring · Repayment recommendation — each a specialized model, not one LLM doing everything

✅ Iterate prompts quickly

Four prompt loops to test

Vendor onboarding questions · Fraud detection follow-ups · Repayment explanation style · Tone in local language vs formal language

✅ Challenge bad outputs

Concrete validation tests

Did AI extract the right bill amount? Is the loan recommendation too high? Is the repayment plan realistic for a daily-income vendor? Does the bot ask fair questions?

✅ Best Use of AI

AI replaces the loan officer

Six AI roles work in sequence to do what a human loan officer does — at WhatsApp scale, in any language, at 11pm

✅ Most Innovative

Alternative Trust Score

No credit bureau. Creditworthiness built entirely from conversational and business signals — serving a market traditional finance can't reach

✅ Most Realistic

Deployable today

WhatsApp Business API exists. OCR and STT are commodity. The regulatory model (NBFC partner) is proven. Real market: 190M+ micro-vendors in India

Prompt Starter Used

"Act as a product manager and senior engineer. Design a WhatsApp-first micro-financing product for small vendors. Identify the user, the core cash-flow problem, and a differentiated AI feature. Show exactly what AI does in the product, then propose a simple technical architecture including data inputs, AI components, and a human review step."

6
Product Name

Why Hisaab (हिसाब)?

हिसाब — the word vendors already own

Hisaab means "account" or "calculation" in Hindi and Urdu. It's the word a vendor uses when they settle with their supplier: "mera hisaab karo" (do my accounting). It implies fair reckoning — not charity, not gatekeeping. The AI is simply doing the math on the vendor's behalf.

  • Resonant for the target user — already in their vocabulary, no translation needed
  • Emotionally accurate — it's about accounting for your business, not receiving a handout
  • Not available as a major fintech brand in India (unlike Dukaan, Vyapar)
  • Tagline writes itself"Aapka hisaab, aapki zubaan." (Your account, your language.)
🏆
Best Use of AI
💡
Most Innovative
💰
Would Fund This

Explore the full product

See how the six AI tools fit together and how the Trust Score works.

View Product → Back to Home