Home > Columns > Executive Interviews

Covisian Executive Interview

Fabio Sattolo, Chief People and Technology Officer, Covisian

Click image below to view the ebook

 ebook image

In this discussion, Sheri Greenhaus, Managing Partner of CrmXchange, speaks with Fabio Sattolo, Chief People and Technology Officer at Covisian. They explore the challenges and innovations surrounding AI integration in customer service. Fabio shares Covisian's approach to combining human expertise with AI solutions, addressing concerns about control and reliability. Together, they shed light on how this hybrid model ensures effective customer interactions and agent support.

Sheri: I wasn't that familiar with the company, but I see the yellow smiley faces all across the conference.

Fabio: We are seeing a lot of people coming to our booths, and this is good because everybody wants to smile. That's what we do. This is part of our strategy. It's not just a communication; it's because we have tried to use technologies, particularly generative AI, to find solutions that help us bring smiles to our customers and consumers. We aim to leverage efficiency and technology to create an easier world.

We don't just use technology because it's trendy. We aim to figure out a model where technology brings real value to people. In our company, we truly integrate technology and people. My position as Chief People and Technology Officer brings these two elements together, which are usually separate in companies. Typically, you don't see IT and human resources together, but we believe that in this period, where technology heavily impacts processes, especially in our sector and contact center space, we need to develop technology and people in the same place and direction.

This is why our company decided to unite both the HR and technology departments under one direction. I believe this is really useful because our technical solutions exploit technology while bringing real value to the human aspect. This approach benefits our clients by lowering costs without losing the human touch in interactions between the client, company, brand, and their customers. Our solution is based on these principles.

We aim to have call centers where, when you call, a human will help you solve your problem. We believe technology and artificial intelligence should be tools to enhance effectiveness and lower costs while providing a brilliant customer experience. Our approach focuses on using technology to bring value to people, not just to cut costs. Efficiency is one of the goals, but it should not be the only one. We believe this kind of technology, especially in our sector, will bring significant value and efficiency.

We believe in sharing this efficiency and value with all stakeholders, particularly the agents who do the work and the final customers who desire a brilliant customer experience. Of course, the brands will also benefit by reducing costs.

Sheri: When you say you're helping the agent, what are you doing? How is it assisting the agent? Are there bots the agent listens to as well, or what's happening with the agent? Explain to me what they're seeing. 

Fabio: We have integrated several technologies, and artificial intelligence is used for three main activities.

First, we create bots, or virtual agents, capable of having conversations—whether written or voice—with the final customer.

Second, artificial intelligence acts as a co-pilot, providing real-time hints and suggestions to the agents. While they are talking, the AI listens to the conversation, reads the knowledge base, and provides suggestions. For instance, it might say, "You're discussing this issue; please refer to these articles and talk about these topics." This helps the agent have the right information without needing to search the knowledge base manually.

Third, AI is used for analytics. It takes the conversation recordings and provides insights into the topics discussed, the sentiment of the conversation, and whether the agent followed the script. It helps ensure the agent has communicated everything necessary to the customer.

We have combined these elements into the agent desktop we are developing and added some specific features, like real-time calculation of the customer experience. We aim to provide an environment where the agent has several tools, such as suggestions, call summaries, and activities provided by AI. Additionally, they have the possibility to use bots as colleagues to delegate part of the activity, let's say you need to collect information from your customer, like their address, phone number, or credit card number. This activity doesn't inherently add value, and there's no need for the agent to handle it. So, we can delegate that to virtual agents, with one important characteristic: the agent maintains control over the interaction with the virtual agent.

The agent watches the transaction and monitors what's happening. We've developed a customer experience index that allows the agent to understand if everything is going well without having to read all the transactions. This index analyzes the interaction between the virtual agent and the customer. If something goes wrong, the agent receives an alert, and the conversation turns red. Then, the agent can take back control of the conversation to ensure the final customer's experience isn't compromised.

If the bot isn't performing properly, the agent can intervene, saying, "I understand the bot isn't meeting your needs. Please talk to me, and I'll try to solve the problem for you." This approach allows us to introduce artificial intelligence responsibly. Otherwise, if the interaction between the virtual agent and the customer goes wrong, there's no one to address it. The customer ends up frustrated, unable to deal with the AI, and may eventually hang up and call for human assistance.

This scenario often occurs when consumers try to reach a brand and encounter a bot that doesn't understand. With our solution, we maintain the efficiency that bots bring while retaining control over the customer experience. Imagine a desktop where the agent can manage multiple calls simultaneously because their colleagues handle the dialogue with customers. The agent has access to the customer experience index, which helps identify calls that aren't going well. If something is amiss, the agent can step in and rectify the situation.

Sheri: What happens if the agent is handling multiple conversations? The bot's dealing with a bunch of things, and now the agent sees the bot's not performing elsewhere. So, does the agent have to stop that conversation to attend to the bot? How does that work?

Fabio: We can configure the bot so that if something goes wrong, it can pause the conversation and say, "My human colleague is busy at the moment. Can you wait, or would you prefer to talk to the next available agent?" It's like when you're put on hold and asked if you'd like to continue with the same agent or switch to another one. Essentially, the artificial intelligence manages this and waits for the agent to become available. It's configurable, and it's up to the customer to decide whether they want to wait for the same agent or be routed to another one. This can all be managed and configured in the system.

Sheri: You mentioned training and agent support. So, does it provide information that's not in your training module? Does it offer space where training can be accessed? Also, about recording and transcription, is it not part of your reporting platform, but you work with recordings?

Fabio: It's all within our platform. We can record and store transactions, transcribe calls, and there's a defined retention period, which can be customized with the customer. All this data is encrypted with private keys for security. We use the public cloud on AWS and provide the private key only to the customer.

We can't access this data because we lack the key; only the customer has it. So, the data is well-protected. However, the recording and transcription are handled by us. Actually, for transcription, we use various market solutions. We keep exploring new ones each week, but we've developed a tool that's platform-independent. This allows us to utilize different services available on the market, not just sticking to one. The same goes for artificial intelligence; we use GPT and others like AWS cloud. We've built an integration layer ensuring our platform's behavior remains consistent, allowing us to use various AI models. If the customer has their own, we can accommodate that.

Sheri: Is this one platform with several modules?

Fabio: Yes, it's a single platform with multiple modules. Today, here in Vegas, we're presenting my CX Pro, basically the agent desktop. It includes all the functionalities to support agents in introducing AI into the process. But there are several other models built to support the customer operation of call centers. These include modules for managing the workforce, planning activity, and defining agent requirements based on the value that needs to be managed.

Sheri: And it looks like there was also a simulation model for workforce planning?

Fabio: Exactly. It's a workforce management tool that helps efficiently manage call centers based on optimal agent scheduling. We've developed an algorithm to allocate people to calls effectively, ensuring we make the best use of agent hours. This model informs team leaders whether specific personnel are needed at specific times, optimizing efficiency in the call center.

Sheri: When speaking with people at the conference, what do you find is the biggest issue they're having?

Fabio: Usually, during this period, everyone's looking for savings. The biggest issue they mention is that while they understand artificial intelligence can lower costs, they're concerned about two main things. First, losing control over their customer relations if they delegate AI to manage them. And second, there's a little problem with generative AI, which is reliability. They don't want AI generating incorrect answers, which is a significant concern. Generative AI works statistically, creating sentences based on probabilities, which can sometimes be wrong. For example, if a customer asks if a specific item is covered by insurance, and it's not in the policy description, AI might still say yes based on statistics, leading to incorrect information.

With our solution, we give customers control, and human agents can verify AI responses. We don't currently use generative AI to directly interact with customers due to privacy concerns. Instead, we use conversational AI to deal with customers, while generative AI calculates customer indexes for agents, who decide whether to use that information.

Sheri: What else would you like our customers to know?

Fabio: They should know this is a responsible way to introduce AI where humans still have control. Many predict that in the future, 80% of calls will be managed by AI. Our solution proposes a different model where 100% of calls are managed by humans and bots together. Imagine calling and immediately getting assistance from a human who then utilizes AI to efficiently solve your problem. It's a better experience because when you call, you usually want to talk to someone. The process can then be managed by the bot, but the human addresses your issue first.

Many times, AI can handle the process itself, but customers may not be comfortable with it, especially those with language barriers. So, the agent can choose not to involve AI for specific customers. Also, if most calls are handled by AI, who will sell to customers? Our solution involves agents in the process, allowing them to solve problems efficiently and then upsell to customers when needed. This model brings efficiency and ensures customer needs are met effectively.