We met with Alexander Kuznetsov, co-founder and CBDO of Voctiv, a developer of voice assistants based on artificial intelligence. The company was founded in 2020 in the United States and now operates worldwide. We discussed with Alexander why businesses need Conversational AI, how to develop such services in fundamentally different markets, and of course, what else virtual agents will learn in the near future.
Let’s talk about Voctiv. What tasks do you help your clients solve, and in which markets do you operate?
Our team creates Conversational AI solutions for call centers that automate routine customer service tasks and improve business efficiency. We have built a full-fledged communication platform that can handle complex dialogues, in multiple languages, across different channels, including voice and text.
We have several products we offer to our customers. First, a Virtual Agent that can handle incoming and outgoing phone calls. Second, a Mobile Personal Assistant for mobile subscribers. Additionally, a virtual secretary Voctiv App, an application that can receive a customer’s calls, answer questions, and even schedule a meeting in the calendar.
These universal products allow Voctiv to operate for a broad range of clients. Our services are beneficial to companies with large call centers or those that require a large volume of calls. We help them optimize customer communication costs and improve their current business processes. Additionally, we attract partners and developers who use our technologies to build their products. We offer them SaaS solutions, training, and support to help them develop their own solutions. We also offer a new service to cellular operators, which replaces voicemail, allowing an increase in voice traffic and subsequent revenue. Our services are also useful for small and medium-sized businesses who do not want to miss important calls from clients. Our AI assistant can record messages, conduct a dialogue, ask questions, and schedule meetings in the calendar to ensure that no important calls are missed.
We develop business almost on all continents: in North America, the Middle East, Europe, Southeast Asia, and North Africa, in total – fifteen countries.
What markets are you responsible for as a CBDO?
I am in charge of business development in all countries except the United States and Mexico. This includes Turkey, Saudi Arabia, Kazakhstan, India, Vietnam, Japan, the Philippines, Indonesia, and others. My responsibilities include interacting with clients, organizing operational processes, pricing, and revenue in general.
How do you manage to build a successful business in countries that differ so much? Are there national differences in how they relate to talking to a virtual agent?
Surprisingly, we don’t see any difference, subscribers all over the world are fine with talking to an AI. I think the main reason is because we make our product so natural. As long as people receive quality service without feeling any difference, they don’t mind conversing with an AI. This was the main idea behind founding Voctiv – to create a service that can imitate human speech, including emotions, complexities of dialogue, intonations, and pauses. Our virtual agents communicate in such a natural way that only 1% of respondents can spot the difference. Most people think they are talking to a real person, which makes us effective even in competitive markets like the Philippines, where the call center segment is well developed.
It seems that customer communications and call centers are a long-understood and established market, voice assistants are also not uncommon. How did the idea come to you to start a business in this particular segment?
Before founding Voctiv, I worked as a product manager in a telecommunications company. During my time there, I witnessed how operators in call centers communicate with customers using scripted responses, which made them like robots. I had an idea that voice bots could be used to talk to customers and respond to them based on pre-defined scripts, allowing human operators to focus on more complex tasks. A few years later, thanks to artificial intelligence, we were finally able to develop a voice assistant that was indistinguishable from a real human.
My partner, Nikolay Kravchuk, who is also a co-founder of Voctiv, had previously developed voice projects in the telecommunications industry. We had worked together in the past and shared a common interest in artificial intelligence and machine learning. Together, we decided to create a company that would help increase business efficiency not just through automation but also through personalized experiences.
From the very beginning, we focused on making our product user-friendly and easily integrated into our clients’ systems. As a result, our Voctiv services quickly became in demand worldwide.
Why are AI-powered speech technologies better than traditional call centers, especially with native speakers?
Our services can greatly enhance the efficiency of call centers. One of the ways we achieve this is through automated dialing. In many Asian countries, even autodialers are not used and individuals enter numbers manually, wasting a lot of time. Our system engages with the subscriber without human intervention, freeing up operators to handle more complex tasks. Our virtual agent follows pre-defined scripts to handle routine tasks. Another benefit of our solution is that it helps to reduce staff turnover, which is often caused by employees getting bored and tired of repeating the same mundane tasks.
A good example is when typical requests are processed in customer service. Often, call center operators can respond to a request using a standard phrase from the instructions or FAQ. Based on our experience, around 80% of customer support requests can be resolved with standard responses. In cases where the answer cannot be found by AI, the remaining 20% are transferred to human operators. This means that call center operators can focus on solving more important and unique problems while the automated system handles routine workloads.
Another difference between traditional call centers and our AI Contact Center is scalability. In case of events or situations where more communication and operators are required, instead of hiring and training new employees, virtual agents can help. These agents are trained to perform at the same level as the best call center employees, and they can be deployed quickly in large numbers to meet the rising demand.
Another great advantage of our service is the ability to speak multiple languages. This is especially useful in countries like Singapore, where there are three official languages. Call centers must employ workers who speak each of these languages. Our platform can now recognize over 160 languages. We have clients working in different countries, so their subscribers may speak Vietnamese, Japanese, and more. When they call our call center, our virtual agent can understand the language and continue the conversation without the need to switch lines or seek another employee.
Voctiv operates in 15 countries, supporting over 160 languages – an impressive feat. Was this capability available at launch, or did you develop the product for communication in just one language?
The primary focus of our project was to develop a natural conversation platform, without any intention of making it international. However, we soon realized that our platform could easily interact in various languages, including complex ones like Arabic, Japanese, or Chinese. Although this was challenging from a technology perspective, our team’s AI expertise made it possible. Even though none of our team members spoke Romanian, we were able to create a voice assistant in the Romanian language by following a simple process. We only require a thousand call records in a specific language to analyze the data, develop the dialog flow, and prepare a dataset for training the NLU (Natural Language Understanding). As we collect more data, we retrain the NLU until it reaches an accuracy of no less than 95%.
Conversational AI is data-driven, and the more data we have, the better the performance and quality of communication.
But many languages are very different, belong to different groups, and have their own intonations and dialects. And what to do with more rare languages or even mixed ones, such as Singaporean Singlish? Does the choice of the market to operate in depend on the complexity of the local language?
Actually, the principles of dialogue are universal: the first phrase, the answer, the analysis of that answer, and the continuation. Our platform enables seamless dialogue building in any language. We have successfully worked with Bahasa, Tagalog, Vietnamese, and Japanese languages and have found that new languages do not pose an obstacle to our service. We recruit specialists, explain how the platform works, and guide them on the production of the script.
However, business operations in different countries present unique challenges. It is crucial to understand the local specifics and have a flexible approach. We adapt our strategies to each market, considering the business culture and its development pace. This includes building strong partnerships locally, understanding the legislation, and exercising patience.
As an expert in the field of voice AI assistants, what do you think about their development in the coming years?
A few years ago, it was predicted 95% of customer interactions would be carried out using artificial intelligence by 2025. We see that these forecasts have not come true. AI is being implemented with caution and attention to how it will impact society and the economy. This technology will certainly find application in new areas such as education and medicine. Voice assistants will be more actively integrated into cars, smart homes, and a host of other devices and services.
But it’s important to understand that they won’t replace people, but rather will free up their time for more interesting and creative tasks. For example, we don’t walk up to the fifteenth floor but take the elevator, and in the same way, we shouldn’t waste time and energy on routine operations.
