Having been at Verint for a number of years, across a range of roles including Engineering, Partner Training and Solution Consultancy, I’ve always had a grasp on the technologies and the structures required for WFM. However, in recent years I’ve also gained a lot of experience of the user side, through working and training with organisations like the Forum.

In a similar vein, I’ve long been aware of and interested in AI and even studied it in my university days – but I’d never have believed where the technology would be today and its influence on the CX world. We all know AI is either the best thing to happen in years or will be the end of humanity – or somewhere in between, depending on who we ask.

The key thing I’ve learnt from customer projects, however, is that they often start with the technology, before the business goals have been defined. For example, do we need to ‘buy some AI’ or do we in fact need to reduce handle times and drive customer and employee experience? How will success be measured, what data do we need, and how disruptive will all of this be?

This is what I now see every day: AI outcomes driving real improvements for our customers, leaving them to focus on what they’re best at – while we worry about the AI model and data used to get there. If you could reduce average call duration from 6 minutes to 3 minutes, would you need to know how the AI did it?

So on that note, let’s purely focus on exploring the business outcomes that are achievable, rather than any analysis of which AI model is best for which task (which isn’t as exciting as it sounds).

When investigating solutions, these are the questions that need answering:

• What automation can it achieve?
• How can it deliver better tools to my employees?
• How can it enrich the data available to us?

Let’s address each of those in turn.

Automation

Automation has always been a big part of WFO, but recently that’s stepped up to include containing contacts, automating tasks such as call summaries, or empowering agents with self-scheduling. Imagine not having to say ‘no’ to agents’ requests to move shifts, instead being able to say ‘yes’ – while also maintaining (and improving) overall schedule health. With the Verint TimeFlex Bot, responses can be changed from a no to a yes, while giving the user clear visibility of what needs to happen for their request to be accepted. Gamification can be used, if appropriate, to incentivize agents to balance business and personal needs.

What’s so great about TimeFlex is that moves are assessed for their potential cost in terms of schedule fit. As long as they can balance out ‘unhelpful’ changes with ‘helpful’ ones, and at least break even, then schedule changes can be accommodated at short notice, avoiding any need for approval.

Aside from the functional advantages, the benefits also cover improvements in areas such as employee NPS (+32% in one case). This comes back to my original point about outcomes – customers can achieve these benefits without extensive data preparation, model selection, or even any knowledge of the underlying AI processes. If schedule fit can be maintained or improved, and agents’ requests fulfilled, where’s the downside?

Better tools

This is the area where the progress of AI has been incredible. It doesn’t seem long ago that just seeing a call being transcribed was mind-blowing! The idea of doing it in real time and using it to guide agents seemed almost unbelievable.

One key theme around AI is that it can often be used to do one thing, but to do it really well – whether that is playing chess, identifying fraudulent bank transactions, or reducing AHT like in the below:

Each of Verint’s CoPilot Bots have one task, and they do it incredibly well. This helps them – and our customers – deliver shorter calls, happier employees and more satisfied customers.

A couple of examples of the CoPilot Bots’ specialisms are:

• Coaching Bot: Provides in-the-moment, non-disruptive guidance to agents, helping make every agent your best agent.
• Interaction Wrap-Up Bot: Uses generative AI to automatically summarize calls and post to your system of record, reducing after-call work.

If you needed a clear idea of how AI can help in the CX industry, this is a pretty powerful one.

Better data

Even 20 or so years ago when I first heard about AI, it was always built on the foundation of data. Indeed, one of the drivers behind the recent explosion of AI is the fact that we now have more data then we’ve ever had before. One AI tool told me the global volume of data is projected to reach 181 zettabytes by the end of 2025. If you've never heard of a zettabyte, don't worry – but just for contrast, in 2005 the total amount of data created, captured, copied, and consumed globally was around 0.1 zettabytes.

Good quality data is essentially the lifeblood of AI. Without it, models cannot be trained to make accurate predictions.

It can be advantageous to look at AI, in very simple terms, as:

• data (which is used for training)
• some kind of training process, and
• an outcome such as a prediction or an automated task.

Google’s AlphaGo was trained on about 30 million expert moves and numerous games, Deep Blue used thousands of chess games, and at Verint we use our Engagement Data Hub:

This seamlessly blends data from multiple sources (e.g., third-party APIs, Verint products, CRM platforms). It’s worth remembering, the bigger the dataset, the more information a bot is trained on – but conversely, the more irrelevant data it may be trained on. If we pointed our bots at the entire internet, there would be a lot of noise in there! Instead, our bots continually train in Data Hub (the Bot Gym) to become more and more effective – there is no degradation over time, as the bots are ‘fed’ with fresh data 24/7. Again, the numbers are mind-blowing – data from over 10 billion real-world interactions annually!

But it’s not just the bots that can take advantage of this rich source of data. Anyone who needs it can easily access its analytics and insights, meaning they can search and discover relationships across all the data within the Hub. No analytics skill or long training courses are required, and our algorithms even auto-surface trends, anomalies and other headlines – just like a helpful, all-seeing assistant letting you know what areas to keep a close eye on. When data is as large as some of the numbers quoted, it’s really the relationships between, and insights into, that data that really matters.

In this article I wanted to showcase some of things I’ve learnt from working with AI daily and seeing how it benefits our customers. The main point, strangely, seems to be to stop talking about AI – and instead, focus on the outcomes. If I found a model which predicted the lottery numbers, I certainly wouldn’t position the fact it uses AI as its main selling point!

Similarly, at Verint, I’ve seen how solutions like our CoPilot Bots deliver up to 3 minutes savings per call, how TimeFlex Bot increases agent satisfaction and improves schedules, and how our Engagement Data Hub can help alert users to potential issues in their operations without the need for a single report to be run.

Hopefully when it comes to AI in CX, this article – and Verint’s cutting-edge approach – will help to reframe the question from ‘how can we invest in AI’ to the far more pertinent ‘what results will we see from it’?

If so, please feel free to reach out to discuss further.

Tom Hunt - Solutions Consultant at Verint.

I really hope you enjoy this article, if you'd like a chat about anything around it, please reach out to me at thomas.hunt@verint.com.