In May 2023, Aadhaar architect and Infosys co-founder Nandan Nilekani predicted that India would soon become the use-case capital of the world, with the AI revolution in the country being voice-led. True to his words, several startups are now capitalising on the growing demand for voice-based conversational AI use cases in India.
Startups like Velocity, a revenue-based financing platform, are moving beyond its core business of lending to launch a conversational AI product specifically made for banks and NBFCs.
“The move is triggered from the use-case need within the organisation and then we felt, why not we can open it up to others as well? We are already in talks with banks and NBFCs and see that there is very good demand for a product like this,” said Abhiroop Medhekar, cofounder and CEO of Velocity.
The conversational AI solution developed in-house by the Bengaluru-based fintech can deliver a human-like, interactive customer experience transforming the way financial institutions interact with their customers.
Conversational AI is artificial intelligence that simulates human conversation using natural language processing (NLP) and machine learning (ML). It enables machines to understand and respond to human language, facilitating verbal or written interactions.
Traditionally, conversational AI consisted of chatbots and virtual assistants. However, with the advent of newer technologies, including GPT 4o (Omni), voice-based conversational AI is becoming popular.
Open AI's GPT-4o is being rolled out in a phased manner and will support over 50 languages, including several Indian languages. The company said that it has significantly optimised token usage for these languages, with reductions of 4.4x for Gujarati, 3.5x for Telugu, 3.3x for Tamil, and 2.9x for both Marathi and Hindi.
Gnani. ai,backed by Samsung Ventures, is yet another startup that provides custom voice AI models for different verticals and has invested in research to improve the accuracy of their models.
"We have multiple patents in voice AI and have custom voice AI models for different verticals...In addition, with voice phishing, security is a concern and Gnani.ai offers a highly accurate speech biometrics software to prevent potential fraud with automated voicebots," said Ganesh Gopalan, Co-Founder & CEO, Gnani.ai.
Also Read: OpenAI announces launch of GPT-4o — and it's free
Enterprise demand for voice innovation
Enterprise customers as well as mid-size firms are now increasingly asking conversational AI startups to build voice-based customer support assistants.
“We are building a voice-based support to one India’s largest textile retail chain. The demand is enormous. Companies today are more open to adapting and implementing newer technologies that save cost,” said a founder on the condition of anonymity.
Reliance Jio-owned Haptik is also seeing increased demand from enterprises.
“With the AI penetration today, every enterprise is keen to implement voice-based customer support functions. As a result the motivation and the demand for this is much higher. Now, we will see a lot of voice-based AI use-cases implemented too,” said Aakrit Vaish, cofounder and CEO of Haptik .
“There is going to be a lot more demand coming from customers. In the next five years, when we are speaking to agents on the telephone we will probably be talking to AI for the most part,” Vaish added.
Haptik is an AI-based conversational platform that supports multiple Indic languages for creating customer service bots and virtual assistants.
Similarly, conversational AI platform Yellow.ai is also working on voice-based LLMs for customer service functions.
“The way voices are being generated and how human-like conversation is being developed is fascinating. We are working on something for customer service functions” the firm’s co-founder and chief product officer Rashid Khan told Moneycontrol.
Yellow is also working to launch large language models (LLMs) in Kannada, Hindi, and Tamil among others, specifically for customer-service functions, the firm’s co-founder and chief product officer Khan told.
Investors bet big on this segment.
AI startup Sarvam AI recently launched its first commercial voice-to-voice endpoint tool which will help businesses across voice-related functions like customer support.
Sarvam was the first Indian AI startup to raise $41 million in its Series A funding round led by Lightspeed Ventures with participation from Peak XV Partners and Khosla Ventures in December of 2023.
“ “We will see lot more players coming into this voice segment. Voice as a natural interface for e.g. customer service functions is already being implemented by a lot of companies in India as well as across the globe. There are a number of AI receptionists as well today. With newer tech like ChatGPT 4o and more of these voice-based AI use-cases will get targeted” said Hemant Mohapatra, Partner at Lightspeed India
Sense AI, a VC firm investing in AI-first startups since 2017 has found that in 2023, 26% of companies in India employed AI for automating customer service interactions and to reduce operational costs. The VC firm said in a whitepaper.
“The expanding capabilities and decreasing costs of AI signal a surge in AI applications, poised to redefine productivity and transform job markets. Research indicates that AI-adopting firms are on track to double their market share, with efficiency gains potentially reaching tenfold by 2024,” Said Rahul Agarwalla, Managing Partner at SenseAI.
Yet another founder said that he is signed atleast three deals in startups that are working on voice-based LLMs and use-cases in India.
“We have invested in around three startups last week working on AI innovation using voice. These startups need funding and training but they have huge opportunity ahead,” the seed-stage investor said requesting anonymity.
Challenges Ahead
Adapting language beyond English, especially Indic language support through voice is still a challenge.
“However, the problem is, that unless startups solve this end-to-end, it is very difficult to succeed. Startups should focus away from the fast commoditizing voice generation layer and solve end-to-end for customer service-related AI functions to survive," Mohapatra of Lightspeed said.
Data also remains a challenge for startups.
"Another practical challenge is regarding data. At the foundational layer, innovations require a lot of investment and synthesizing of real-world data which is easier for those with large capital pools. Innovation is fast moving upstream towards the application & solution layers," Mohapatra added.
Similarly, for startups to implement voice-based customer support functions within large enterprises may be challenging.
“We have spent several years to built something with voice however never executed something end-to-end, beyond technology challenge for voice for customer service, implementation in large enterprise set up takes a lot of time and effort. It is also sometimes not ROI positive However integration and implementation challenges within large enterprises still remain. We have to see how this pans out,” said Jio Haptik’s Vaish.
However, players like Bhashini and Sarvam are working to bridge this gap.
In March IIT Madras’ research lab AI4Bharat on March 6 launched IndicVoices, an open-source natural and speech dataset, covering 22 Indian languages.
"Unlike text-based chatbots, voice-based chatbots require deep tech and custom AI models – in terms of speech-to-text accuracy, understanding multilingual conversations, speech activity detection...Speech to text accuracy is not a completely solved problem and making the voices in voice-based chatbots or voicebots more human-sounding is another Gopalan of Gnani.ai added.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
