For several years now, business experts have advised corporate leaders to place artificial intelligence (AI) at the heart of their strategic thinking. McKinsey says that it is essential for ‘pulling insights from oceans of data.’ Forbes says that ‘AI is going to transform every business, in every industry.’

But how will AI really change every business? What if your business has not fundamentally changed in decades or more? What is new with AI, and especially the generative AI that has captured recent headlines? Perhaps a real example from an energy company may indicate where we are heading.


Octopus Energy is a UK-based European energy supplier focused on delivering sustainable gas and electricity. The company has 3.2 million customers and the British Consumer’s Association has consistently named them the most recommended energy supplier for the past six years.

In February 2023, they plugged ChatGPT into their customer service system and allowed it to start replying to customer emails. Customer satisfaction with emails written by humans was rated at about 65%. The same measurement for those written by AI was over 80%.

By May 2023, Octopus was answering 34% of all their customer customer emails with AI. That’s the equivalent of 250 human customer service professionals writing those replies.


Greg Jackson, the CEO of Octopus Energy, commented on this experiment with AI in an article published in The Times: “Today, Octopus AI writes messages, but soon (in weeks) it’ll make decisions and carry out actions too. It won’t just tell you your balance but will also suggest a change in payments and do it for you. It’ll order your meter installation or spot a vulnerability and send you an electric blanket.”

Consider what this means for your business?

What if the simple bots we know today, that tell you where your parcel currently is or how much you owe your credit card company, evolve to listen to requests and then make decisions and take actions. The bot could even suggest ideas and solutions to help a customer without ever needing to refer to a human supervisor.

But let’s step back for a moment. What is generative AI really good at? It can take a large body of data, search for relevant information, connect the ideas together and generate a meaningful answer or comment.

Imagine how this will change the work of paralegals that need to search for case precedents. Imagine how this could help doctors to diagnose illness based on a series of test results. Imagine how software developers can automate vast chunks of their coding tasks.

The Octopus Energy case study is a great example of a situation that is perfect for a generative AI system. There is a large body of knowledge available to train the chatbot – both general knowledge as well as detailed knowledge of the products offered by Octopus. There is a written question or request from the customer. The interaction is not in real-time. The AI can generate a text response and email it back to the customer and there is no need to wait for feedback.

Now think about the complexity of live human to human communication. There are five steps when one human communicates with another: idea formation, encoding, channel selection, decoding and feedback. In short, one person needs to think about what they want to say, form a sentence, choose whether they will say it, text it, or email it, the recipient needs to understand the message and respond.

It’s an extremely complex process that we all take for granted. When we hear the first words from a baby it is the result of that child hearing the same sound thousands of times and realizing that they have the ability to recreate it. Eventually they attach meaning to the sound and it become a word. Then emotion can be attached to the word as it is said.

If you search Google for information about the complexity of human communication it will return over 89 million articles and books. It would be far easier to ask ChatGPT to summarize all of them.

In our area of expertise, the customer journey and improving customer experience, there have been many predictions around how generative AI will change the business. Some have predicted the death of the contact center itself. Almost all the commentators are wrong.

I’ve seen how our business is changing. We are deploying generative AI solutions in many different areas, including chatbots and voice bots that can directly interact with customers. Email can be automatically parsed and understood – as in the Octopus Energy example, but it’s also easy to do it in any language because machine translation can be applied to the text.

Inside the contact center we are already using AI to monitor the quality of customer interactions. Every single call or message can be monitored now, but just a random sample. Speech itself can be analyzed to provide insight into how the customer was feeling during a call.

Complex interactions still require a human. There will be many processes where executives find that AI can be helpful. It can summarize large reports, seek trends in that ocean of data, and it can help to improve and automate many customer service processes.

This is clear inside Webhelp. We are already deploying solutions that facilitate multilingual customer interactions, that can summarize a customer interaction for our quality systems, that can coach agents by focusing on weaker customer interactions and offering guidance, and creating language models that include detailed product information.

Just imagine a chatbot that not only understands clear human language when you ask it a question, but it also knows everything about the product in question because it has absorbed every document every written about the product inside the data used to train it.

All of our experience inside Webhelp and that of other companies, such as Octopus Energy, is demonstrating that there is a genuine business case for the use of AI. It can communicate with customers.

However, the Octopus case study also demonstrates that the best way to use this technology for customer interaction is where there is a break in the middle – the communication is asynchronous. This allows a brief period where the AI can analyze and formulate an intelligent response to a WhatsApp message or email.

In conversation this is not possible. AI bots that utilize a voice synthesizer and Natural Language Processing (NLP) are becoming very good. You can now have a natural conversation with a bot, but because it has to be trained on a body of knowledge it will not be able to anticipate the correct response to an unusual situation.

If a customer calls an energy company to explain why they have not paid their bill it would be much harder to identify a vulnerable customer in real-time by listening to their words, compared to reading a written email.

AI offers a great opportunity to automate many different business processes. It will affect every company in every industry, but for the majority of human-centered processes, people are the most valuable asset your business can possess.

At Webhelp, we strongly believe that AI will have a very limited impact on complex conversations, which currently represent the vast majority of the time advisors spend actually interacting with customers. The impact is even more limited in that the rate of adoption of AI by businesses and end customers (after all, not everyone wants to interact with a bot!) will be significantly less than 100%. All our analysis clearly demonstrates that the human, boosted by AI, is the present and the future of customer experience.