1 of 15

Multilingual Voice based AA Consent

​

Reverie + Accion for

Sahamati BuildAAthon 2024

2 of 15

The Challenge

  • Theme - Risk Management solutions for the ecosystem

​

  • Digital Lending has brought a revolution in terms of lowering total cost per transaction, enabling penetration into low ticket size loans.

​

  • AA journeys are not language inclusive for the digitally challenged strata of the society. This presents a challenge- the Loan journey completion vs Understanding of Consent with the Borrower is a very thin line to balance and achieve.

​

  • Account Aggregator journeys see a significant drop off at the Consent stage due to various factors, Some of them include-
    • Consent is not understood and its Cognitive load during the loan journey is quite daunting, resulting in drop offs.
    • User journeys are not localised or language/dialect personalized to provide clarifications to the borrowers.
    • Assisted journeys include Consent stage that run into additional risks in terms of borrowers not even being made aware of the consent effectively.

3 of 15

Multilingual Voice based Consent

for Agent assisted Journey

4 of 15

Solution Approach

  • AA Consent is collected over a Phone Call. The Phone call is driven by a Gen AI Bot that-
    • Ask the User all necessary information to complete the consent journey.
    • Answer any clarifying questions that the user/borrower may have.

​

  • The scope of this solution scope is for Agent assisted journeys Only. This scope is restricted for the Hackathon, and can be extended to Self-serve journey if needed.

​

  • As part of the Hackathon, the team has built an AA Adapter, which sits on top of current AA participant, and triggers the Voice call which communicates with the User, Accept Consent feedback and respond back to the AA participant. Once the Consent feedback is taken from user the AA participant returns the control back to the FIU app.

​

  • The AA Adapter is built on top of Saafe AA APIs and is fully integrated with Sahamati Sandbox environment.

5 of 15

Participants in the Solution

  • FIU - Trigger the Consent to AA (Reverie+Saafe) and receive customers feedback in either Accepted or Rejected.

​

  • AA Backend (Only Saafe APIs) - Will participate in the network and interact with other AA participants such as FIUs. It will also serve as backend for Call flow to send consent dynamic data (linked accounts, purpose, amount etc) needed during the call flow.

​

  • Voice Bot - Will drive the Consent conversation with the end User, requires dynamic data (linked accounts, purpose, amount etc)

​

  • Reverie AA Adapter - The idea of the adapter is to act as a glue between Voice bot and AA Backend. It will be a small application but will need to have both frontend and backend
    • AA Adapter Backend - Wraps several calls to AA backend so that Voice bot API tools/triggers are lightweight. This will also trigger the Voice calls, and receive final feedback from the user and will pass it back into the AA Backend. It will also serve the AA Adapter frontend.
    • AA Adapter Frontend - a single page application, mainly used to host a page that is a redirect_url when the Consent is initiated from the FIU. Additionally, it also provides breakdown on each sub stage during consent feedback session.

6 of 15

Participants in the Solution

7 of 15

Voice call based Consent Demo

Live Trial - link (please use 1234 as OTP)

8 of 15

  • Conduct PoC in the field partnering with AAs and validate usage of taking Consent on Voice calls.
  • Integrate the AA Adapter front-end with Assisted Journey so that Agents can track progress on the application during the IVR Calls.
  • Consider possible Integration with WhatsApp so that Borrowers can track Consents given in a simpler manner.
  • Extend Multi-lingual support for IVR and Whatsapp for all Indic languages.

Future Scope

9 of 15

Multilingual Consent

for Self-serve Journey

10 of 15

Self-Serve Demo

​

Apart from Voice call based Consent, the team has created a PoC for Do-It-Yourself or Self-Serve Journey that is Multilingual & Voice enabled for AA mobile app. Please see demo-

11 of 15

Participating Organizations

    • Accion - User Research, AA Advisory, Bot Script design
    • Reverie - Gen AI Voice Bot Integration, AA Ecosystem Integration, Indic Voice AI tech

Participating team

Extended team

Reverie + Accion Team

12 of 15

Accion

Accion is a global nonprofit on a mission to create a fair and inclusive economy for people who are failed by the global financial system. We develop and scale responsible digital financial solutions for small business owners, smallholder farmers, and women, so they can make informed decisions. Through targeted investment strategies, advisory solutions, and expert thought leadership, we work with local partners to develop and scale affordable, accessible and customer-friendly financial solutions. Since 1961, Accion has helped build 267 financial service providers in emerging markets of 75 countries, reaching 440 million people. More at https://www.accion.org.

​

Accion Advisory combines decades of on-the-ground experience with insights into new technologies to help our partners become more sustainable and scalable.

13 of 15

Accion

Understand barriers to use AA tools

and data

Identify & test

use cases for

prioritized segments

Pilots to test

use cases for self-served and assisted channels

Pathways to scale the account aggregator usage

Our team provides strategic and operational support to strengthen our partners and maximize their impact through:

  • Expansive fintech knowledge and reach
  • Deep experience in designing, launching and optimizing digital products, channels, and strategies
  • A team of 50 people with a vast geographic presence, with strong understanding of local context and customer needs
  • Expertise in using data analytics for risk, credit, and customer strategy
  • Hands-on experience as technical and managerial advisors

Working to improve penetration of Account Aggregator services with the support of Gates foundation for the underbanked population in India with the following objectives in focus

14 of 15

Proud Part of

Reliance JIO

​

As a division of Reliance JIO Platforms, we draw on its technological prowess and innovation to deliver superior service

​

Decade-Long Proven Innovation

​

Our 13+ years of history, with 200+ members in R&D, is filled with consistent innovation in Al and language localization, highlighting our commitment to technological advancement

​

​

​

Future-Ready Solutions

​

Our Al ML solutions, ready-to-deploy, far exceed the tender requirements and are scalable to meet any future localization needs

​

Indic Languages trained NLU

​

​

Unlike other models which are trained only on English, Reverie's model is

trained on all 22 Indian languages

​

Reverie - Commitment Meets Innovation

WHY

REVERIE

Jio Backed Robust Financial Stability

​

​

Backed by Reliance JIO, our financial stability guarantees dependable, uninterrupted service delivery for the project's duration

​

​

Made in India In-House Tech Solutions

​

​

Our in-house indigenously developed Al ML solutions allow us full control over our technology, ensuring rapid adaptation to evolving needs

​

​

​

​

Prestigious

Clientele

​

​

We have a proven track record of servicing a wide spectrum of prestigious clients in the private and government sectors, including Parliament of India, Income Tax, MHRD, myGov, JIO, Tata Motors, and many more

​

​

Only Global

Player

​

​

Only entity offering end-to-end Al ML language localization solutions across 22 languages for any platform or format

​

​

15 of 15

Reach us at

www.reverieinc.com