AiThority Interview with Babak Pahlavan, Founder and CEO of NinjaTech AI

Most organizations are capitalizing on Artificial Intelligence (AI) capabilities to automate repetitive tasks to save time and resources. By 2035, nearly 90% of global organizations would rely on AI tools for their customer service management. This would boost their organizational productivity by 40 percent, and increase revenues by up to 25 percent. To meet the global demands of customer service and support, it is important to keep a track of the AI innovations, particularly in the conversational AI and text generation fields, that directly influence AI-assisted customer service plans in the industry. In our exclusive AI-centric AiThority interview series, we are hosting industry leaders in customer service to identify the opportunities, challenges, and innovations needed to scale the existing AI-driven CX strategies. Today, we have NinjaTech AI’s CEO and founder, Babak Pahlavan.

This is what Babak said about AI’s growing dominance in customer support industry.

Why is an AI-assisted customer service plan better than human-only efforts?

Babak: An AI-assisted customer service plan has several unique components that make it compelling:

Breadth, accuracy and speed of knowledge

AI enables the ability to answer a broader list of customer service questions with increased accuracy. This is accomplished by utilizing and training AI on a “mixture of experts” model. This means that depending on the query or customer support question needing to be answered, an AI agent is able to draw on multiple sources of information — including various LLMs, Clouds and Search — and place the highest confidence interval around the potential answers. Best of all, this breadth and accuracy of knowledge is computed in milliseconds, so to a customer it is seamless and instant. In a human-only customer service model, the breadth of knowledge extends only to what the customer service agent has been provided by the employer, the accuracy beyond that limited knowledge set is limited, and the speed of response will be considerably slower than an AI-generated response.

Breadth of interaction channels

Due to cost and volume factors, most companies tend to push customers towards chat and email channels.

With AI, all channels can be available all of the time with instant access — an AI agent doesn’t fatigue, and it can handle almost limitless interactions simultaneously. This offers the opportunity to dramatically improve the customer experience by offering customers the channel of their preference — whether that be video with an AI avatar, phone call with the AI, chatting or emailing. Being able to select the preferred interaction channel enhances the customer experience and improves customer satisfaction.

Economics and insights at scale

There are considerable economic advantages to using AI — in particular the ability to bifurcate tasks that can be handled by an AI versus those that may require human intervention. With Generative AI, the vast majority of customer service queries can be handled by a well-trained AI with the right LLM access and decision engines. This means lower cost of service, reduced ticket volume, and faster resolution of tickets.

Equally, AI only becomes better as it ingests data and insights. With the regular use of AI in customer service, the model and decision engines are reinforced with the learning of each interaction and the models become better — able to not only handle each customer service call better, but to also generate macro insights on the types of issues / support queries customers are bringing. These insights can have impactful benefits on product development, company strategy and long-term capital investment.

How can contact center owners transform their customer experience strategy with Conversational AI?

Babak:

Start small

Incorporating Conversational AI into workflows is not an instant thing, it can be disruptive to customers and the support workforce is not done well. To avoid implementation friction, consider starting with one channel (eg. chat) and focus it on solving a very specific set of queries that can be trained with a high confidence interval. Once this use case proves out, begin to expand into other use cases in the same channel. Over time, branch into multiple use cases and then multiple channels.

Test multiple third parties

Building your own LLM and decision-engines is hard — this requires chops in deep tech. There are numerous customer service providers emerging that are building their own ‘GPT’ architecture to tackle the customer service challenge. In-fact, it is likely to be one of the most crowded spaces in the coming years as venture funding for younger companies and capital investment with more experienced companies accelerates. Consider trying a number of 3Ps on trial and benchmark their performance against each other to pick a winner. The accuracy, latency and cost will be important metrics to evaluate as you consider the best 3P to go with.

Eat your own dogfood

It is often easy to consider customer service / support to be the end of the customer experience value chain — but it is critical to long-term customer retention. Set the bar for an AI and human experience at a level that you personally — the CEO and Product leadership of the company — would expect and test it heartily. At Google, our product and engineering leadership teams would often spend times in our call centers picking up phones and absorbing the customer experience to ensure they were touching the customer experience at every end of the value chain. Introducing AI will be a meaningful change to the customer experience, apply a high bar and test it to ensure it is tweaked and massaged to meet that bar.

Thank you, Babak! That was fun and we hope to see you back on AiThority.com soon.