Lately, enterprise executives have been asking me a lot about Artificial Intelligence (AI). What is it? why there is so much hype about it? And what can we tell our stakeholders about it? After explaining and discussing that this is no hype and it can profoundly transform the way employees work, the next big question is, how can we start using it?
Beyond the benefits that come from telling your board and investors that the company is using AI, other benefits include detecting fraud, predicting customer behavior, automating recurring tasks and reducing operational costs, while providing the best customer service, to name a few.
I’ll start by focusing on how companies can introduce AI in a pragmatic way into their daily processes. Forrester Research recently described Pragmatic AI as “comprised of discrete technologies that are advanced enough to add intelligence to customer service and deliver quantifiable value.”
This post is the first in a series, where I will explore the many benefits and best practices of implementing Pragmatic AI.
Let’s begin with AI-Based Application Programming Interfaces (APIs).
AI has been used for many years by consumer products companies and we, as individuals, benefit from it every day when we do a Google search, or when a photo is tagged by Facebook, or when Amazon recommends a product for us.
These innovations are based on R&D that cost hundreds of thousands of dollars and are part of the intellectual property of their respective companies. But there are open APIs that enterprises and developers can benefit from. Existing applications can become smart solutions by integrating AI functions like recommenders, natural processing languages and document processing algorithms.
Let’s explore a concrete example related to customer service in companies that use Interactive Voice Response (IVR) technology and record conversations. An automated process can be created that can review the hundreds of daily conversations, identify the customer sentiment and classify the type of problem based on language. Let’s breakdown the technology behind it:
- Convert the recorded audio to text using one of the many speech-to-text APIs (Google Cloud API, IBM Watson Speech to Text, Bing Speech API)
- Run a sentiment analysis per conversation to get a satisfaction score (MS Text Analytics API, TheySay)
- Record results and analyze against demographic data
Not only can the enterprise identify all incidents quickly, but they can address them and increase customer satisfaction by measuring customer health and reducing costs. It also frees up agents from routine tasks and allows them to focus more on interactions that require deeper analysis. Customers are starting to expect personalized and intelligent experiences and AI-based customer service apps do just that.
A proof of concept (POC) or experiment, can be quickly done to measure the benefits of specific AI solutions within a week using different sentiment analysis APIs, and you can then select the one that best suits your needs.
As you can see, AI is in reach for any enterprise, as we have seen in this customer service scenario, and in many industries.
In the next post, I will take this to the next level by discussing how chatbots, supported by Natural Language Processing, Computer Vision and predictive Machine Learning models, can interact with customers in a variety of settings.
I would love to hear your thoughts on integrating AI into the enterprise. For future blogs, I’ll be exploring new topics, such as AI’s cultural acceptance, educating staff, and how AI is being applied in specific industries. I look forward to your feedback and new areas of focus are always welcome!