AI will be the next step growth factor for cloud telephony: Ozonetel CIO

For cloud computing-based telephony services provider Ozonetel, the Covid-19 pandemic brought in new business. Among other things, the Hyderabad based company witnessed a 100% growth from February 2020 to June 2021, in both revenue and agent number terms. It signed on its first $1 million deal and forged Covid-19 related triaging facilitation partnerships with several state governments. 

The company, founded in 2007, found itself in a perfect place to grab new opportunities as clients, operating across spaces of food delivery, online pharmacies, or even legacy banking, paced up their migration from an office space-required customer support set-ups to remote ones.  

In an interview with TechCircle, Chaitanya Chokkareddy, the chief innovation officer at Ozonetel, spoke about the recent growth at the company, scaling capabilities of the technology stack and rise in cloud telephony business in India. 

Edited Excerpts:  

How has Ozonetel’s business grown over the last few years?

We started back in 2007. 2011 was when we launched our current primary product, the cloud agent, which is our cloud call center product.  

Since 2011, we have been pitching cloud as a solution for call centres. Slowly, it's been picking up over the last six, seven years. In the initial stages from 2011-17, mainly startups were our customers because they try new things, instead of going with the old-school call center technologies. They were the likes of BigBasket, Zomato, and Practo. Since 2018, we have been seeing enterprises, and even old school businesses adapting cloud.  

But 2019, and then when Covid started, it became completely mainstream. Call centers had to run operations, and the only way they could run it was with cloud, because there is no way especially with lockdown that the agents can come to a common location. Rental was the major cost for contact center operation. And so, they used to pack these humans together.  

After about two months of complete shutdown, food services, health care services started opening up, and they had to take support calls. But all agents were back in hometowns. What also helped was that the government was proactive in restricting certain regulations, such as other service provider (OSC) licenses. When you have run a contact centre operation, you have to take a whole host of licenses and for particular locations. They removed that specific clause, and that allowed contact centers to be remote. 

How did these macro-economic conditions affect Ozonetel’s business?

It has contributed to more than 100% growth (from February 2020 to June 2021) --that is from a revenue perspective, as well as number of agents’ perspective. We sort of doubled those kinds of things. We got our first $1 million customer also, but I'm not able to share the name as it’s an ongoing process. We are working with partners through this, which we have started from 2018-19. 

How was the competition in new deal sign-ups, given cloud telephony demand?

Our competitors’ cloud offerings are more of add-ons to their existing on-premise offerings. They're not pure cloud, so to say. We are pure cloud, and that is what actually attracted the bank towards us. The advantage with the cloud is that anything that a customer asks, we can deploy immediately, within like a week. That kind of a turnaround time, the banking and insurance segments had not seen before.

How have Ozonetel’s healthcare sector services grown?

From a healthcare perspective, if you look at it, the communication has always been core. One of our first customers was the Practo, who ended up using our software to connect doctors to patients and stuff. This was seven, eight years before anything happened. From them onwards, we have lot of other healthcare clients like Dr Lal Path Labs and Cloudnine. 

 We have been in the space for a long time. But when Covid hit, it became again a priority, though we didn't proactively reach out initially, from a sales perspective. At that time, practically speaking, we were not thinking about sales, we were thinking survival from a personal basis… StepOne, a non-profit, was launched by Raghavendra (Prasad T S), Qikwell that was sold to Practo in 2015, and others. They had reached out to us and other cloud telephony operators. They asked for setting up of a pool of cloud telephony resources and help governments handle calls for medical triaging.  

We worked with them and set it up for multiple state governments – Maharashtra, Odisha, Assam – this was during the first wave, along with a lot of other cloud technical barriers, we were able to set it up very fast.  

During the second wave peak in April-May 2021, we worked with our partners and enabled Covid related state help lines in states like National Capital Region (NCR), Meghalaya, Andhra Pradesh, Karnataka, Tamil Nadu, Punjab, Maharashtra among others… we enabled many of these help line contact centres to be up and running within 48 hours.  

Subsequently, all other healthcare customers that we had before wanted the sudden increase in capacity. They would have to buy the server, and put it up in that. All of that would be dead inventory for them. Because of cloud, they are on demand, clients able to scale up and down, depending on the pandemic’s waves. In a traditional set-up, that would have never been possible. 

How are these contracts modeled? 

It is mostly monthly, majority of our contracts are monthly, but we have been moving. Enterprise customers generally do yearly, annual contracts. 

How was Ozonetel prepared for the scale of businesses around the pandemic? 

That worked out for us, mainly because of the architecture we had chosen 10 years back, when we set it up itself. For a horizontal scaling, a microservicing architecture that we had set it up for, when we initially installed the system.  

Obviously, we didn't plan for this, nobody planned for Covid. But when it happened, and we were asked to sort of scale… if you look at it from 10 years before, to now, even data centres have scaled. They have become very good at managing their infra. 10 years back, it was a huge problem for us, because we had to call them personally, talk to them. Now everything happens over track. 

How have your data centres relationships to vendors evolved? 

Initially, when we started off, we used to rent complete server racks, and manage our own servers. That was the only relationship between the data centres. So, all the relationship we had 10 years back with the data centre operator was that he will make sure that network connectivity and power was provided to the rack, nothing else. And we managed everything.  

But slowly over time, what has happened, in fact the biggest change that has happened is that AWS and Google Cloud have set up their infra within India. In Mumbai, they have set up their own access points.  

This wasn’t the case earlier. And, for us being a telephony service provider, we couldn't run on Amazon AWS or Google Cloud, if it was outside of the Indian shores because all calls should land and all the data of calls should be handled within India. That’s a regulation problem.  

So now that AWS and Google Cloud have come within the Indian shores, we were able to move a lot of our operations to AWS, rather than managing our own racks. Those even now, we still maintain.  

Reliance and Sify are our two main data centre operations. We manage our own lot of servers, but a majority of central operations, under Disaster Recovery (DR) sort of moved to the AWS. And that helped because when the sudden scale was needed, that we were able to get scale on Amazon by just adding virtual machines.  

How does Ozonetel view the future of the cloud telephony business? 

I think almost all domains and not only contact centres, are being encapsulated by Artificial Intelligence (AI). Even for us, that's the next step growth factor. Even from a product perspective, as well as scale perspective, we have been concentrating on sort of augmenting our contact centre software with AI. This especially comes from the fact that artificial intelligence has grown leaps and bounds in the last couple of years, especially in the Natural Language Processing (NLP).  

If you look at, for example, open AIs GPT-3, which sort of generates text almost as equivalent to a human. And that sort of helps us to help our agents come up with the right kind of sentences to respond to people. So, this we are integrating with our chat products. A lot of AI is going into our chat. Along with calls, we also provide an omnichannel contact center support. So, we handle chat, we handle email, and we handle calls.  

A lot of development is going on towards our chat product development. Every contact centre wants to move to chat, for a simple reason that one agent can handle 10 chats at the same time. But one agent can handle only one call at a time. So calls wise, it's always a big cost for a contact centre.  

Unfortunately, chat has its own problems in the sense that many people are not comfortable chatting, even now, though, it will change in with the millennials, they sort of want to chat more than talk. But when it comes to support, they rather prefer to talk they would rather just pick up the call and tell the problem and keep it off.  

Chat also requires multilingual support.  On the other hand, with calls as long as the other person understands it, we are fine. So we are trying to work through this, with translation systems… Finding a good chat agent is hard in Telugu, not because they can't speak Telugu, it's because there are no good input equipment or keyboards. Then we have a good transliteration engine where you can type in English, and gives out Telugu characters. That happens very little on mobiles. Agents are on desktops, mostly… We are not deploying anything in production right now, because the translation output is not dependable completely, yet.  

Similarly, the other areas from a speech perspective, we have humongous amounts of call data available. We handle almost like 8 million calls per day. Each call on an average turns out to be around two minutes if I take so that is like 16 million minutes of calls a day I'm handling right now. So, this kind of data is available with us.  

We want to work and see whether we can come up with speech analytics. Just by listening to the frequency pitch and the way you speak, loudness, can we predict whether a customer is angry? And if the customer is angry, can I give the right input to the manager, so that the manager or supervisor can barge in to the call and pacify the customer. So, these are some tools which are in our pipeline, we are already experimenting with this with certain customers, beta customers. We identify some four or five customers and we deploy these for them. These are the growth areas for us in the coming year.  

How is the cloud telephony stack modeled? 

In fact, when we started, we used to build our own Primary Rate Interface (PRI) cards, which connects to the telco, we used to build the hardware ourselves. On top of that is our telephony stack, which we call it the cookie stack. And on top of that is the cloud agent, which is our application stack. Now what is happening is on top of this application layer is where the AI layer is coming in. It’s already built for cloud. The advantage is that all I need to do is use all the data generated by all the stack below us, and feed this to our training agents of AI.  

Could you tell us about your Google solutions integration? 

That basically is one more channel. What happens is we need to be available where customers are reaching out to businesses, or businesses reach out to customers. What are the different channels? Traditionally, it has been voice, maybe SMS, and email. Email and voice have been most used… these channels are becoming old school. Now people are reaching out to businesses and businesses are reaching out to me on multiple other channels.  

This includes Facebook, WhatsApp, Instagram, and Twitter. That social layer has started. If you observe, now, who has lost out on all this? It is Google, because people search on Google, but they close the transaction on the website.  

The chat happens on the website, or the call happens on the website. Now what Google has done is they've added a new product, where when you search, from the search page itself, you can directly chat with a business. So they've added this new channel, last year in beta mode in the United States. It’s part of Google Business Messages solution. They are coming out with multiple products inside of that. It comprises of Google Business Chat, Google Verified Calls, and Google Ad messaging or chat which is yet to be released… this is going to be a huge thing. Not a lot of businesses have been exposed to it yet. We were when we were one of the first people in India to pick it up. 

What does this partnership mean?  

No, it's not like that. These are open APIs. They had just invited some people to use, because these are very innovative and new APIs. Now I think it’s open to all. Last year, they opened to only a few sets, we were also called because of the number of calls we handle. Actually, they reached out to us for Google Verified Business Calls. But as part of that, they provided access to this also, Google Business Chat, where we integrated with that, and we provided all these features. For example, if you search for Ozonetel on Chrome in Mobile, you can directly start a chat with my agent from the Google search page itself.  

What’s the business case here? 

What will happen in the coming year, is that beside that call, there will be a chat button.  

Currently, Google drives humongous amounts of traffic, either through web search, or through calls also. In fact, from their own prediction for one Pizza delivery operation in the US, they have driver driven almost 60 million calls in one year. Now, as I explained to you, calls are a big problem for businesses, because agents have to handle those 60 million calls. So, businesses are trying to drive traffic towards chat. Now beside the call option, you have a chat… I think this particular product is going to go places for Google, because it will drive lot of chat traffic too.

Chatbots, unfortunately, have not picked up. They were promised as the Holy Grail in 2016 but then they sort of died down. Nobody uses chatbots, per se. It works only for small concerns like finding order IDs, and then it generally, it falls back to an agent. I think that's where even our growth, will see a lot more chat agents.  

Speaking about growth, how do Ozonetel’s business expansion plans look? 

Product wise, I think we have a pretty stable contact centre product. We are adding more channels, like the Google one. And that's more of a horizontal thing, but a vertical stack wise, we will be adding the AI piece. From a sales perspective, this year we are going to mainly concentrate in the US. So last year, we sort of realized that US was a big market for us. We thought contact centre space has saturated there, but then when we launched, we figured out that there was space there. We already crossed $1 million ARR in the region, last year. This year, we want to target $5 million ARR in the US. India business as it is we grow. Here, it is a market size problem, because the market is picking up now. As and when the banks and insurance segment and other BPOs, start picking up cloud contact center, we will grow here normally. The good thing is we have a name here and I think we will continue to grow in a nice straight line. 

What strategies will drive the growth in the United States business? 

For that, we will try to replicate what we've done so far. We will be going mostly through a partner route. We are identifying partners in the US who will be selling our product. That is an ongoing process. We have already identified four partners. They’ve have been selling it so far for now. We will be identifying more parties this year and we hope to grab the $5 million ARR business. 

How about the South Asia, and EMEA businesses? 

We are already there in Singapore and UAE, and that is going similar to India. I mean, that's more of a linear fashion growth for us, because we have sort of established our presence there. And we keep on growing regularly, mostly through inbound.