How Reverie and Boonbox create desi robots

The global chatbot market is expected to reach a valuation of USD 102.29 billion by 2026, registering a CAGR of 34.75% over the forecast period 2021-2026. Efficiency. According to the MIT Technology Study, up to 90% of companies reported faster complaint resolution with bots.

Advances in natural language processing (NLP) have made chatbots more accurate, intuitive and autonomous. However, most of the chatbots available in the market are English first or English only. According to statista, only 1.35 billion of the world’s population speak English as their first language or as a second language.

Since 2009, Reverie Language Technologies has offered content in multiple languages ​​in real time to bridge the language gap. Reverie has now branched out into voice-driven social commerce activations and conversational AI for contact center automation. Recently, the company partnered with Boonbox (a proprietary rural assisted commerce platform) to cater to the needs of the non-English speaking Indian community.

In an exclusive interview with Analytics India Magazine, Vivekananda Pani, Co-Founder and CTO, Reverie Language Technologies and Ramachandran Ramanathan, CEO, Boonbox talked about their plans to build multilingual bots from scratch.

AIM: How does Reverie build multilingual chatbots from scratch?

Vivekananda Pani: We consult with companies to understand “why the bot is needed”, “what languages ​​are needed”, “are users comfortable typing in their language?”

Then we understand that use cases need to be automated through the bot. This is a step-by-step process to understand the desired user journeys for the different actions the BOT is supposed to perform. Broad compartments for the same can be order or transactional. Once we have the desired flow plan, we (along with our partner) convert the existing process into a conversation flow between the end user and the bot.

Once the flow is finalized, the deployment is directly worked on the chatbot platform. This bot development includes API integrations with any of the third-party APIs. The development conversion of the bot flow would take into consideration the usage behaviors of an Indian-speaking user and modify the flow accordingly. Once the bot flows are developed, intents and entities are then identified for Natural Language Understanding (NLU) training. It’s about creating all possible variations in which a user could ask in their language.

For example: a user’s intent may be to “check bank account balance”, in this case the user may ask the query as follows:

  • English:
    • “What is my account balance”
    • “How much money is in my account”
  • Hindi
    • “मेरे खाते में कितने पैसे हैं”
    • “मेरा बैंक बैलेंस कितना है”
    • Simple bank account mein kitna balance hai

NLU’s training in Indian languages ​​is what gives our technology the edge. It is trained on the native script, so the NLU is able to better understand the context and provide greater accuracy. Text-to-speech (STT) is also fine-tuned based on the use case to provide the highest accuracy. Once all components are ready, automated tests are performed on all components to ensure stability and performance. Another part of the testing is done by the language users to ensure that the bot is able to answer the different questions from the users. Failing cases are then recycled using a human-in-the-loop approach to make the engines and bot robust.

Finally, the bot can be published seamlessly on a wide range of channels such as IVR lines, website, mobile apps, WhatsApp, Facebook Messenger and many more.

AIM: How does the Whatsapp multilingual bot work?

Vivekananda Pani: A WhatsApp bot is like another contact on your phone: either the bot can initiate the conversation, or you can start just by saying “hi” or simple commands like “do you have any table fans” (in the case of a social commerce interface). At the very first step, the user is asked about their language preferences, the user can either give input via menu based buttons, type the message and even by voice. Users can always change the language at any time during the conversation. Once the user begins to interact with the bot, the message begins to pass through Reverie’s voice suite to deliver responses in the selected language, either by text or voice notes.

AIM: What is the technology stack behind your products? What languages ​​does Reverie support?

Vivekananda Pani: Reverie’s Voice Suite supports English and 10 Indian languages ​​including Hindi, Marathi, Gujarati, Bengali, Tamil, Telugu, Kannada, Malayalam, Punjabi and Odia.

The technology stack:

  • STT/ASR: Reverie’s Speech to Text (STT) module is developed using advanced deep learning technologies and carefully selected data with coverage of various accents, ages and genders.
  • Language detection: Language detection is an essential part of the multilingual chatbot, because we need to understand which languages ​​are spoken by the user before they can be understood by the NLU engine.
  • NLU: Reverie’s Natural Language Understanding (NLU) is developed using state-of-the-art transformer technologies that understand Indian languages ​​as well as English.
  • Reverie’s TTS: Text-To-Speech provides fast and natural human speech in both genders with extensive customization options to generate perfect responses.

AIM: Which industries are seeing the highest demand for multilingual bots?

Vivekananda Pani: The main areas where we see the highest demands include:


  • Core Banking – To check account balance, transfer funds, mini statement, checkbook request
  • Collections- Loan collections for different loan products through multiple channels like IVR, WhatsApp.
  • Customer Support – Product Information, ATM/Branch details
  • For employees – Retrieving information from banking process documents for compliance etc.
  • Analyze – agent sentiment, tone and performance.


  • Complete the customer onboarding journey, including payments – Purchase life and general insurance (health, travel, auto, etc.)
  • Service Requests – Policy renewals, change of nominee, address, claims handling, policy, etc.
  • Analyze – agent sentiment, tone and performance.


  • Product Ordering – To select and order products through multiple channels
  • All transaction commands


  • Order and search for use cases like – open, launch or play/view a program.
  • Language and channel switching use cases.

The most demanded channels for the same are IVR bots, voice activated chatbot and Whatsapp.

AIM: How important is interacting in the local language to build customer trust?

Ramachandran Ramanathan: Boonbox has sold products to over 2 million customers in rural India. Our customers are confident, aware and eager to improve their lifestyle. Boonbox is completely “local” and our interactions with our customers are only in the local language. Our clients are from different states and one of the common characteristics is that they do not converse or communicate in English. Conversing with them in their local language means increased interaction and in doing so builds trust. Whether it is verbal communication or via applications, the language is, by default, local.

AIM: What is the scope and potential of “voice commands” in relation to rural and social e-commerce, especially via Whatsapp?

Ramachandran Ramanathan: The voice is intuitive, making it an ideal means of communication with any group of customers. Broadcasting certain messages to customer phones has always been part of Boonbox’s go-to-market strategy. Overall, most customers in non-metro cities are uncomfortable typing even if they have had some school education. It’s disconcerting that companies have yet to realize the potential of voice in customer interactions. The key here is to make the customer understand that voice communication is an option. Boonbox connects with its customers and also drives business transactions using Whatsapp, where the use of Voice could potentially be a game-changer.

Vivekananda Pani: Over the years, WhatsApp has become a familiar customer communication channel and is establishing itself in the era of social commerce and Direct to Customer (D2C) players.

AIM: What does the partnership with Reverie mean for Boonbox?

Ramachandran Ramanathan: Reverie’s expertise in Indian language technologies and solutions makes it an ideal solution for Boonbox, where “local” is the default option and the only way to build customer trust. Reverie’s role would cover Boonbox apps, communicating via Whatsapp and using voice technology whenever possible. The Boonbox-Reverie collaboration has the potential to make e-commerce and social commerce second nature to Boonbox’s rural footprint.

Comments are closed.