Artificial intelligence (AI) is revolutionising customer service and reshaping the way companies engage with their clientele. Using AI-first automation makes use of deep learning, machine learning, and natural language processing to improve scalability, efficiency, and personalisation. Artificial intelligence (AI)-driven chatbots and virtual assistants offer round-the-clock assistance, manage large amounts of queries, lower operating expenses, and enhance customer satisfaction and response times. Predictive analytics and sentiment analysis help to better customise consumer encounters. AI-first automation offers several advantages over human intervention, including increased customer loyalty, cost savings, and increased efficiency. However, ethical concerns and the loss of human touch are some of the issues this technology faces.
This week we were joined by Co-founder and Chief Executive Officer of Yellow.ai, Raghu Ravinutala. Yellow.ai, established in 2016 with a vision to revolutionize customer and enterprise interactions, making them more effective and effortless. Under his leadership, Yellow.ai has rapidly expanded its footprint across North America, Asia-Pacific, Europe, the Middle East, Africa, and Latin America, serving over 1100 customers in 85+ countries. Raghu’s entrepreneurial journey is rooted in his extensive 20 years experience in the tech industry, where he held key leadership roles in Engineering, Product Management, and Business Development at Texas Instruments and Broadcom in both the Bay Area and India. His contributions to the industry have been widely recognized, with accolades including being named one of the Top 50 SaaS CEOs by The Software Report in 2023 and the CX Leader of the Year at the CX Awards in 2022.
The interview delves into the company’s origins, the inspiration behind its name, and its mission to transform customer and enterprise interactions through AI. It explores the evolution of Yellow.ai from leveraging messaging platforms like WhatsApp to deploying advanced AI solutions that enhance customer service efficiency and satisfaction. Raghu discusses the significant impact of AI on the industry, the balance between automation and human empathy, and the future landscape of customer service jobs, emphasizing the shift towards AI system management and the broader implications for enterprises and consumers.
AIM Research: Why the name Yellow?
“One of the reasons is, we want to make people ask this question”
Raghu Ravinutala: One of the reasons is we want to make people ask this question. It is more remembered when it is quirky. But there’s also a deeper reason because when we started the company, we aspired to change how consumers connected with businesses. Historically they were connecting using yellow pages and now they’re connecting with AI. That’s the genesis of the name. And it turned out to be memorable for folks and is considered cool in the world
of enterprise software filled with blue in general.
AIM Research: Could you walk us through the journey of your thought process leading up to the decision to solve the customer-centricity problem?
“If people were messaging so frequently with friends then they would want to use the same medium to connect with businesses and make it easier.”
Raghu Ravinutala: Genesis was essentially that I grew up in India, and the company started in Bangalore. For anyone who has lived in India around 2010 – 2015 one thing that anybody would have used WhatsApp. It was picking up and had unprecedented adoptions and we felt that the future of consumer business communication will happen a lot on messaging platforms. If people were messaging so frequently with friends then they would want to use the same medium to connect with businesses and make it easier. That was the genesis of the company. Then we launched an application that helped consumers message with businesses and without much marketing it took off. Lot of people started using it to connect to message businesses. A lot of these interactions were very similar. People were asking for appointments or loan status etc. Then we figured why don’t we use AI to automate and reduce the burden on the customer service teams. And then we were very fortunate to work with some pioneering customers in India. I believe India has the most risk taking customers. We worked with Asian Paints, Bajaj finance, all these were our initial customers, and when they launched these AI agents on their websites, they immediately saw the number of calls to their contact centers dropped, higher conversion on their loan books, getting more appointments at Asian Paints and so on. So that was like a product market fit and then we started evolving the product to not just provide information but to enable transactions from chat to voice and machine learning to large language models currently. So that’s been the evolution of the technology industry.
AIM Research: What are some of the key drivers that large enterprises are seeking in terms of opportunities for their customer strategy? And how does Yellow.ai’s ability to enable an AI-first approach toward customer success align with these needs?
“Even with that spend, people are on hold over customer support calls or they are not getting answers really fast.”
Raghu Ravinutala: I think the problem of the industry is indicated in some really large numbers. Globally every year there is close to 1 trillion dollars of customer service spread across companies. Even with that spend, people are on hold over customer support calls or they are not getting answers really fast. So it’s a really big problem because a lot of spending is happening and the needle is not moving in terms of customer service. We believe that AI is ideally suited to solve this really big problem, because if you imagine what is the largest AI application with the most traction, it is ChatGPT. People like interactive chat and companies are providing that automated interface and providing great delight to customers. They are getting answers and questions are resolved really fast and you are reducing your cost of operating your customer service. Typically, companies are spending a lot on outsourcing this customer service as well. So it’s not like they are taking away jobs from their payrolls. Jobs are changing everywhere but it’s not giving companies the political hurdle of doing cost reduction. So it’s a combination of all the three that are coming together to drive this adoption.
AIM Research: When do you think we’ll reach a point where we can completely rely on an AI chatbot to engage in seamless conversation, whether through speech or text?
“With generative AI now, the responses and interactions are a lot more empathetic; they cover a lot more depth and breadth as well.”
Raghu Ravinutala: I always use the analogy of Tesla. I use Autosteer FSD to some extent and when all this auto driving started it started off with just a lane control. The automation was not able to help in changing lanes etc. But it was still useful because there’s autosteer for a long time and now we are getting very close to end-to-end self driving. So the journey of automation of customer service has a lot of analogy with that where it started off with, providing information and whenever there is a complex query, very fast product humans move to helping enable transactions, but it was not empathetic. With generative AI now, the responses and interactions are a lot more empathetic; they cover a lot more depth and breadth as well. In the future, I think this will continue to get to an extent where it becomes fully autonomous where AI can help make decisions and reason with the customers to demonstrate human-like empathy. It’s a journey, we are not yet there but with the current technology, you can still go ahead and automate 60 to 80% of customer service interactions, for 60 to 80% of the customer service queries. You can have phenomenal satisfaction rates. Because what it covers for the lack of depth of emotions and empathy in conversations it covers it with speed and accuracy. We call our contacts and we want to get a thing done or just reschedule or ask how the weather is etc. You get it done in two minutes. But if you are worried about COVID protocols, or some serious issues that you want to talk about empathetically like humans, it is not yet there and it doesn’t convey that feeling yet to the customers and that’s when humans are needed, for 60 to 80% of the tasks.
AIM Research: What key research or innovation is needed to enable AI to better understand and interpret complex human emotions and nuances like sarcasm, and what advancements is Yellow.ai currently making in this area?
“We are a company that solves customer service, and the best we can solve in that narrow domain is what we work on.”
Raghu Ravinutala: If you see the large language models currently, they’re solving just one problem mathematically which is predicting the next best word from all the training of the language that they have which by itself is not recognising emotions. So it is kind of correlating for the given context of what is the next best word. But what is also coming together is models that are large visual models where you’re combining this large language with all the vision data so you’re monitoring that potentially in this use case of human interaction.But at the same time, we need to be clear what kind of company we are. We are not a company that is focusing on how AI can be human-like in all possible ways. We are a company that solves customer service, and the best we can solve in that narrow domain is what we work on. For example, we don’t need to really work on how AI needs to respond. We are not there but there is AI software that is specifically focused on that. There’s AI software that is helping in mental health. We’re not that. We are a company that is focused on how fast we can get the customer problems resolved. How fast and better we can communicate when there is a problem with the company etc. And now the data set becomes much much narrower where there are historical conversations that we can train on and have our system do the best possible conversation in a customer service context.
AIM Research: In customer service or hospitality, handling situations smoothly is easy, but how can we effectively manage unhappy or dissatisfied customers by incorporating not just speed but also genuine empathy?
“You need to basically make sure that AI agents for customer service are trained to respond”
Raghu Ravinutala: I’m not saying we should not explore empathy, but what I’m saying is that the whole range of emotions that you need to train a model for is very focused on a customer service use case where the problem statement is very narrow and the data for training is already available because people are already responding. And you can understand from that data on what needs to be the right response and interaction versus solving the emotion problem as a whole on how I react to a tragic thing or a romantic proposal. You don’t need to solve all that. The emotions are wide ranging. But just imagine what customer service agents are getting trained for at this point. You need to basically make sure that AI agents for customer service are trained to respond. You know significantly better than what humans are really different. Because they have a lot more data but you don’t and we as a company are not in the space of solving how well you know or how well human computers can emotionally react. We are in the space of solving for or how best we can respond in a customer service environment.
AIM Research: When it comes to AI-first approaches, common use cases include resolving tickets, identifying root causes, and pulling up data. Can you share an innovative or out-of-the-box use case that has recently emerged, or provide an anecdote where AI has demonstrated exceptional capabilities?
“There are multiple serious use cases of customer service.”
Raghu Ravinutala: There are multiple serious use cases of customer service. One of the things that I really liked and people loved from day one, it was one of our initial customers. It’s a large beverage maker, Diageo. They launched a bartender assistant. Essentially people can interact with Simi, who is a bartender and talk about the ingredients they have at their home, they have at their disposal, and give cocktail recipes. And the cool part is that based on the interactions, the company was able to realize that a certain part of the country preferred spicier cocktails versus a northern part of the country preferred more sweeter cocktails. The way that they advertise their brand and talk about the brand and what they promote is different in those different regions, which they never get the consumer data. But it was a use case that got a tonne of customers for us because everyone felt that it was really nice and quirky. That’s one of the funnier and nice cooler use cases. The most serious ones have a customer in the US called Ferrellgas and they provide propane delivery during winters for heating etc. And their order volumes are quite elastic as you can imagine you might have a lot of gas deliveries for the potential summer versus winter is the peak season. You cannot start the contact center so elastically. Suddenly I don’t want to hire people just for December and Jan and reduce the number of people in the rest of the seasons. Now they have an AI assistant. Now they don’t miss any single order because they’re taking orders through an AI system that’s completely elastic and this draws up their volume of orders working on fulfillment, which is a direct impact on revenue.
AIM Research: What are some immediate advancements in AI that are set to change the way customers interact with companies, and how do you see the role of companies like Yellow.ai evolving over time to help enterprises adapt to these changes?
“What has changed with companies like Yellow.AI is that software is not just storing customer information to help humans provide support but the software itself is managing the relationship and customer support.”
Raghu Ravinutala: I have very strong viewpoints on this. The history of enterprise software and customer service software has been on the premise that humans are providing one on one customer support. And that’s why there are tickets and you go one by one on the tickets and you talk to the customer to resolve their queries. So all the software is designed with that premise. What has changed with companies like Yellow.AI is that software is not just storing customer information to help humans provide support but the software itself is managing the relationship and customer support. Then ideally the support that the support agents are not handling customers one on one, but they are handling customer bases that is one is to hundred or one is to two hundred thousand because they monitor a lot of interactions happening, jumping to the interaction where it’s not working really well, or train the AI versus their core job being handling calls on by on what’s your reschedule or what’s the ticket status answering that. That just fundamentally changes what an enterprise software stack should look like. And we always say that if there is a CRM that’s designed in the AI world, that would look like we are developing. It is a software that manages customer relationships. So we believe that it will evolve in that direction and that propels our ambition and aspiration that we become the core customer service suite for enterprises and replace the historical service clouds and CRMs that have been there for support because they’ve been designed for a completely different world that we are living in.
AIM Research: What are the implications for the job landscape given the advancements in technology by companies like Yellow.ai and other AI service providers? While automation enhances business efficiency, it also generates new roles focused on integration. What is your perspective on the future job landscape and how to approach it?
“So what AI is changing is that it’s making that cost really low to manage the engagement or interaction with the customer.”
Raghu Ravinutala: One thing is for sure, is the way the customer service agents industry that we are working in and 20 million plus customer service agents that are employed by companies. They will see a significant growth in their careers on the contrary of what people are thinking that they’re just going to be eliminated. Here are two reasons that are going to happen. One, if you see any customer company support site where they list the toll free number or way to contact them, it’s somewhere hidden in some form because companies don’t want you to contact them. It’s very expensive for them to support customers. It’s not easy to get their contact number. But companies are also losing the opportunity to interact with their customers and they want to talk to the customers but it’s expensive. So what AI is changing is that it’s making that cost really low to manage the engagement or interaction with the customer. So now companies can go and say
Hey, call us anytime, we are here. We want to increase our engagement rates instead of having 1 million calls per year, the cost is really low, they can have 10 million calls per year. So which means that overall volume of engagement I believe will increase with AI and the agents will transform from handling one on one customer support to more like this AI System Managers where instead of handling 100 customers per day, they probably have 10,000 customers per day. But there is always more about how we train the AI system, how to track it, how do we audit it, monitor it and how do I enter into the interactions that are not going great and need my intervention and things like that. So the job will evolve from a customer service agent to an AI System Manager, AI Customer Service Manager handling from 100 to 10000 clients. So that’s an uplift for them and it’s uplift for companies that are more available for the customers. For consumers as well they don’t need to wait on the phones.