In part one of this blog series, we looked at the idea of creating a Super Agent with the help of artificial intelligence (AI) and utilizing AI to assist agents while mitigating the risks that can come with AI, including legal and societal issues. In part two, we continue exploring how AI can help agents and how contact centers can reduce complexity and operationalize AI.

Keeping Humans in the Middle

Keeping humans in the middle means that the customer is not exposed to AI at all. If the AI hallucinates or comes back with the wrong information, there is an agent there to course correct by rephrasing or finding the accurate information. So, keeping that human in the middle can result in a significant risk reduction for when you’re trying to deal with customers.

This helps to empower agents. It reduces their time to competency, makes sure they’re getting the information they need, helps them resolve issues faster, and gives them the ability to do their job better. It creates Super Agents.

In the contact center industry where turnover is high, giving agents cool, new tools can be fun for them and great for morale. In turn, it provides better customer service because they are dealing with agents that are informed and providing fast resolutions. Then, you achieve throughput because people are waiting less time for an agent, allowing the contact center to deal with a higher volume of interactions.

Reduce Complexity & Operationalize AI

The complexity around AI comes in operationalizing AI in the contact center. Integrating your desktop system with a generative AI application is great, but how will you present the AI to the agent? How do you ensure agents have the information they need, that they can see what they need to see, while also reducing technological risks.

You need to be able to take the AI and connect it with your existing systems including CRMs or a legacy system like AS400, which people still use.

If AI is going to get information about me (as a customer), it must interface with backend systems, and someone needs to figure out how those backend systems work with either RPA or the training of the AI to get it functional. Then, work out a clean way of presenting the details to an agent that is simple and easy to use, so they can provide the information back to their customers.

There is also closed AI versus open AI to consider. Open AI is much harder to make work. A lot of contact center manufacturers are coming up with their own closed AI applications, which automate transcriptions, summarization, intent analysis, and sentiment analysis. In this scenario, the AI doesn’t need to be trained. It’s a product that you can install and have working right away, which helps to lower the bar around operationalizing AI in the contact center.

If you’re integrating AI on the backend or using RPA, that’s more open and is better suited for larger projects. Even with a chatbot, it will need to be trained, and you will have to put guardrails in place so that it understands your industry and terms. This requires a lot of work and there is also a risk of exposure to your customers.

Another challenge is a lack of focus. While pilots are useful, it’s important to focus on one thing at a time, test, fix any issues, and then evolve the pilot so it’s useful. That is when it should get rolled out into production. Having focus will give you concrete results.

AI & the Agent Desktop

Agents need a unified desktop that allows them to deal with any channel, from a voice call to a WhatsApp or a Facebook Messenger conversation. AI will give you the ability to provide information back to those channels, but it won’t bring those channels into your contact center.

It’s important to focus on simplicity and consistency. AI will only add complexity if you don’t have the proper infrastructure to support it, which means having an integrated channel experience. AI can layer on top of your desktop and integrated channels to provide additional value and power your Super Agents, improving first contact resolution. This will also help you track and optimize the customer journey and use an AI application without having to retrain hundreds or thousands of agents.

The Future of AI in Contact Centers

The guardrails we mentioned earlier are situational and will continue to evolve. As will the people who are trying to hack those guardrails. Unfortunately, we can’t pull hallucination out of generative AI, hallucination is part of what it does and the technology to fix it doesn’t exist yet.

When you implement a chatbot, you have it deal with the questions that you get a lot of, that are easy to solve and have high first contact resolution. With these simple tasks, there won’t be a lot of hallucination happening, but it can have a big impact on your agents.

How do we make the contact center agent’s job more enjoyable for them and good for business at the same time?  Reducing the agent desktop complexity and keeping the human in the middle will be important. With AI, there will be less training time, faster time to competency, and the opportunity to boost employee loyalty.