The application of AI is not limited to just customer-facing chatbots and it is also being used for behind-the-scenes applications in areas such as personalised recommendations, marketing, summarization and content generation. In a global perspective, however, Indian startups have not been pioneers, unlike, say, Swedish unicorn Klarna, which used the technology to replace 700 human agents, said Rohit Pandharkar, AI partner, EY.AI is no longer the preserve of just AI firms. Some of India's biggest and most popular startups across sectors like ecommerce and fintech are deploying AI. About two-thirds of India's top 50 most valued unicorn startups are already using artificial intelligence (AI) or generative AI technology, highlighting that startups have been more proactive in terms of adopting AI technology compared to traditional players, a recent report by EY found.
The application of AI is not limited to just customer-facing chatbots and it is also being used for behind-the-scenes applications in areas such as personalised recommendations, marketing, summarization and content generation. It is also being used for fraud detection and risk management, payments and payroll solutions, and more.
"There is no poverty of ambition or talent in Indian startups trying to leverage AI," said Rohit Pandharkar, AI partner, EY. The scale of deployment by Indian startups as well as maturity in AI spends is however still evolving, and still behind US or China, Pandharkar said, adding that this will pick up as the costs of AI go down over time.
In a global perspective, however, Indian startups have not been pioneers, unlike, say, Swedish unicorn Klarna, which used the technology to replace 700 human agents, he said.
"Non-AI-first startups are also integrating AI into their operations, indicating that the distinction between AI and non-AI startups is becoming less pronounced as AI becomes a ubiquitous part of the technological landscape," said Rajnil Mallik, partner and GenAI go-to-market leader at PwC.
Mallik said that while Indian startups looking to adopt AI have the talent pool advantage, global startups gain from having easier access to capital.
On the fintech front, Indian startups may have a leg-up compared to global counterparts, given that the UPI revolution opened up new payments vistas for them to experiment.
ET brings you voices from among India's top startups on how they are solving problems using the disruptive technology to stay ahead in the game.
Flipkart
Jeyandran Venugopal, chief product and technology officer
A critical challenge for users of Flipkart is discovering products that fit their requirements.
This is especially true for first-time online shoppers or those residing in tier-two and three cities, where exposure to diverse shopping experiences might be limited.
GenAI is a lever to improve process efficiency and customer satisfaction and the company is deploying it in over 20 use cases, Jeyandran Venugopal, chief product and technology officer of Flipkart said.
Its GenAI-powered virtual assistant Flippi is a personal shopping buddy that customers can interact with conversationally. You can ask it queries that are semantically very different from what you would put in a search bar - for example, it understands the implicit needs in the statement "I want a mobile phonefor my grandparent", said Venugopal.
The Walmart-owned etailer has also used GenAI to improve the product page, for summarising reviews, enriching product descriptions and images, etc. The immerse feature StyleMate, built on multimodal semantic indexing technology, mingles image and a text description to refine the search. On the seller side, GenAI acts as a virtual customer account manager that sellers can chat with to identify product impressions on the marketplace, get insights and suggestions for greater visibility, pricing and promotions, etc.
"Internally, we deploy GenAI to assist customer support people - at the end, we use GenAI to summarise the interaction and identify if it left the customer happy or unhappy," Venugopal said.
"We are now trying to leverage GenAI to make the vernacular version of the platform even better, providing broader coverage of products and reviews being made available in all languages," he added.
Myntra
Raghu Krishnananda, chief product and technology officer
Given how competitive and crowded the online fashion retail space is, a personalized and engaging shopping experience can give firms like Myntra an edge over their competitors. Considering that its user base is very diverse, and includes the tech-savvy Gen Z and millennial shoppers, Myntra feels that traditional search-based systems might sometimes lead to decision fatigue. Traditional search may struggle to meet the nuanced expectations of these consumers, who demand seamless, intuitive, and highly tailored interactions.
To address this issue, the company integrated GenAI and other emerging technologies across the platform.
"Our flagship conversational AI assistant Maya, powered by OpenAI's ChatGPT, refined user queries. It understands specific customer needs, and offers relevant recommendations from our extensive catalogue," said Raghu Krishnananda, chief product and technology officer at Myntra.
This conversational approach significantly reduced decision fatigue, providing a customized shopping experience, Krishnananda added.
The online fashion space is currently very crowded and while Myntra maintains its leadership in online fashion, Ajio has been a challenger to the Bengaluru-based firm which is owned by Flipkart.
The start-up also launched My Stylist, an AI-driven personal style assistant developed in-house. This leverages machine learning, computer vision algorithms, and browsing data to offer personalized outfit recommendations. This next-generation product discovery tool has high engagement and click-through rates.
MyFashionGPT, powered by ChatGPT, allows users to search for outfits using natural language, offering up to three ensemble options across multiple categories, said Krishnananda.
Policybazaar.com
Saurabh Tiwari, CTO
The financial services sector, especially insurance, is prone to significant fraud risks. Insurance is often considered a low trust category, so even a 1% fraud risk is detrimental to the entire industry. Policybaazaar.com uses a "human-in the-loop" methodology in its risk management framework, where AI and GenAI models support quality and underwriting teams in decision-making.
AI helps identify and flag anomalies at the initial sourcing stage. By examining payment methods and devices, AI filters out unusual activities. It looks at browsing habits and personal data to spot suspicious behaviour and uses knowledge graphs to reveal hidden connections and fraud networks.
"We also use it for face and voice recognition and liveliness tests to detect deepfakes to prevent identity fraud," said Saurabh Tiwari, CTO, Policybazaar.com. Its generative AI models, developed with frameworks like Llama 3 and Mistral, analyse customer-advisor calls. They highlight discrepancies between call disclosures and proposal forms, prompting further investigation by quality teams.
Policybazaar.com's AI-driven risk management framework has resulted in a 95% reduction in fraud, said Tiwari. "It has helped us bring down the total turnaround time from two days to one day for risk assessment. These tangible outcomes highlight the potential for even deeper and stronger collaboration between AI and human teams in the future," he added.
Razorpay
Murali Brahmadesam, CTO and head of engineering
Razorpay, the $7.5 billion digital payments firm with its registered office in the US is using AI to help solve problems for its customers. AI is coming in handy in prioritizing customer pain points and developing tailored solutions that simplify their financial operations.
The Bengaluru-based fintech is applying AI and machine learning to enhance its services, ensuring businesses, irrespective of their size and scale, receive the most effective and efficient solutions available. Its GenAI chatbot, Ray, for instance, is specifically designed for Payments and Payroll solutions for businesses.
The assistant works seamlessly with WhatsApp, web bots, and other channels making it highly accessible and user friendly. It receives questions, retrieves information relevant to businesses from the data bank, and responds to them at the right time and in an accurate manner, the company says.
"Ray is invaluable for businesses, helping them better understand and manage their data, thus improving the efficiency of their operations," Murali Brahmadesam, CTO & head of engineering at Razorpay said.
