Skip to content
Amogh Suman

A voice assistant for borrowers in five languages

A proposal and working demo of an AI voice assistant that lets microfinance borrowers ask about loans, repayments and grievances in their own language, prepared for a microfinance industry body.

Sector
Microfinance
Scope
Proposal and demo, five languages
My role
Solution design and demo build

The problem

Many microfinance borrowers are more comfortable speaking than reading, and prefer their own language. Text apps and English call scripts leave them dependent on a loan officer for simple answers.

The approach

  1. Start from borrower questions

    Collect the questions borrowers actually ask: dues, repayment dates, interest, and how to raise a complaint.
  2. Build a working demo

    Combine speech recognition, a language model and speech output to answer those questions in five languages.
  3. Design the guardrails

    Limit answers to verified information, hand off to a human when unsure, and log conversations for review.
  4. Write the proposal

    Set out rollout phases, cost and how member institutions could plug in.

Who it serves

  • Borrowers

    Answers in their own language, at any hour.
  • Member lenders

    Fewer routine calls for field staff.
  • The industry body

    A shared channel for grievance and financial literacy.

The outcome

A working five-language demo and a rollout proposal presented to the industry body.