Amazon : Flipkart, Uber : Ola, Air B&B : Oyo
Ramakrishna Prasad Nori (RK)
Founder - Head AI Research & Solutions | November 1, 2024
What did the PE firms look for while investing in Flipkart, Ola and Oyo – was it technology innovation, market size or understanding of the Indian market?
There was definitely no technology innovation. That’s obvious. Over the years, I have raised this question in my personal groups consisting of people who understand business and technology & sometimes both. They all agreed. This question had receded to the background for some time but resurfaced again, as we launch hear@ai- a SaaS product.
hear@ai, enables voice recording in Indic languages and English. It transcribes audio into English. It uses non-API based LLMs to analyze transcription and extracts all the valuable insights. More importantly, it can orchestrate action by integrating into client systems, helps close loop the issues with customers. All of this is voice enabled. hear@ai is focussed on the customer engagement and business operations.
hear@ai goes live on May 1st and we have signed up 4 clients and the sales process is catching up good speed. Wherever presented, the response has been very positive and clients are showing interest to take it forward, very quickly.
The product is completely built with the promoter’s funds. For future growth, funds have been raised successfully from friends and family and they are definitely not fools.
How different is hear@ai from other products in its category? This question has come up during our many conversations these past few months with clients and investors.
One straight answer is there is no such product. We (foolishly) tried that as a response the first two times when we heard the question. The response was bewildered looks.
So, we changed tactics. We now talk a little bit more about the AI space even at the risk of glazed looks. There are companies who specialize in training Automatic Speech Recognition (ASR) models and offer it as a service.
There are other companies who train Large Language Models, some charge for usage and some release for free. We test all the models rigorously using real time data. We measure the output and compare the output versus multiple models. We have evaluated the impact of compute cost. We have created a platform so that enterprises don’t have to worry about model effectiveness vs cost vs scalability vs maintenance. For some tasks (from an AI perspective) this is good and sustainable.
When we looked at ways to integrate voice for customer feedback or business operations, we chose a smart way to integrate various methods to solve the problem.
Looking at the ever changing landscape of AI and the problems it can solve, one can choose to use the existing models intelligently and in a cost effective way that enterprises feel confident of accepting the solution as an effective means to solve their problem.
The need for finetuning an LLM with smaller GPUs is possible using a technique called Parameter Efficient Training (PEFT). While I will post more of this later, the need for higher compute is entirely dependent on the degree of intelligence that one is expecting from an LLM. Some tasks will require this capability. While we can train models from scratch and fine tune models, the constraint here is availability of data and access to expensive GPUs.
A breakthrough technology innovation or idea happens once in many years. But start-ups find ways to use the innovation to build downstream solutions to address a problem. Technology can be adapted. If there is a market, the solution creates value and the customers also see the value and when executed well, there will be success. Whether its 1x, 2x or 10x depends on many factors.
It will be nice to know what the investors were thinking when they signed their first cheque for Flipkart, Ola and Oyo.
To prove my point, I tried retrieving an interview with Ashish Gupta (unable to trace the source with Google search) on why he invested in Flipkart. Asked ChatGPT for help and it came back with this answer:
Ashish Gupta highlighted the potential he saw in Flipkart due to its founders' vision, their understanding of the Indian market, and their execution capabilities. He emphasized that Sachin and Binny Bansal had a strong grasp of e-commerce dynamics and were able to effectively execute their plans. Gupta believed that Flipkart had the potential to become a significant player in the Indian e-commerce landscape, which led him to introduce the Bansals to Tiger Global for investment consideration.