For those of us who have been in telecom, AI isn’t anything new. We have been implementing AI features in telecom since the 1970s. Call center automation has been around for 40+ years and could well be considered AI. What is new and what is making AI so important now is Generative AI and Machine Learning. Until recently, AI and automation were great, but every step of the process had to be defined. From what someone did to what each response would be. Every scenario was spelled out with a specific targeted answer. Now, through generative AI, the system can learn and improve its responses as it goes. It can read documents, manuals, etc., and find the answer it thinks is correct even if it must pull from multiple sources. This self-learning and improving process with limited human intervention is what is new and why this is such a big deal. We no longer have to spell out every answer and every response.

When we look at practical applications in telecom, let’s focus on contact centers since they are likely to have some of the most beneficial ROI. You have two types of contact centers: one that is a cost center (tech support helpline) and another that is a profit center (taking incoming orders for clients). Both have very practical uses for AI.

One of the best use cases for AI is Agent Assist. Imagine someone calling into tech support with an issue and explaining the entire problem to your tech support agent. Maybe this is a problem that has happened before, but this agent hasn’t experienced it. The AI solution can listen live to the call and integrate into your CRM system. AI listens to the conversation and looks through past tickets for a match. If it finds something relevant, it presents it to the agent via a side screen. Instead of that agent wasting time troubleshooting from scratch, they get a quick suggestion from AI on what the problem might be and how another agent resolved it. This practical use case can increase customer satisfaction and decrease your average handle time with clients, all while increasing your overall profitability. Yes, you can do sentiment analysis and all sorts of other fancy AI listening tools, but the purpose of this article is to give you real-world cases that have ROI.

The most significant AI functions in telecom are likely to be in contact centers and customer service. However, a common misconception is that AI works out of the box. AI requires accurate, up-to-date content to be effective. Managing this content is crucial—it’s a dedicated job, not something to add to an IT person’s existing workload. The success of AI in your business hinges on content management. Outdated or inaccurate content can lead to mistrust in AI’s suggestions, undermining its effectiveness. This cannot be overstated. Is the content accurate? Does it have access to the latest brochures? Was the outdated content removed when the new information was uploaded? Content management is going to be the biggest gotcha with AI. It will require a dedicated role and cannot be assigned to an IT person who is already managing your network. Yes, AI can increase the overall productivity of your agents, but there is a hidden expense of someone managing AI with constant training and content management.

This brings us to controlling AI to be the most helpful to your agents. The best way to do that is to manage the content it has access to. A simple method is setting up a shared drive with all relevant documents—brochures, product manuals, and specific instructions. This ensures AI has the correct information to provide helpful answers. This is a very basic way of training AI to give answers that work for you and your business. The shared folder can contain special instructions, brochures, product manuals, and everything it would need to answer questions related to your business. Congratulations! We have just solved the most basic level problem of AI: controlling its content.

However, content isn’t static. Keeping it updated is a continuous task. Outdated manuals or brochures must be removed to ensure AI provides accurate information. Failure to maintain up-to-date content can lead to AI becoming less reliable, ultimately reducing its usefulness. The practical use of AI is 100% dependent on the content it has access to. Your worst-case scenario is you spend all this money implementing AI for your business, and your staff loses faith in its answers and disregards its assistance. At that point, AI will stop learning and become more and more inaccurate.

Another practical use of AI is for automatic call escalation, word mining, and tone detection. AI can listen to all calls and look for keywords or tones to escalate a call to a supervisor. Though this is a good practical use of AI, the ROI is ambiguous at best. If you use AI to look for specific words like ‘upgrade my service’ or ‘I would like to speak to a manager,’ these types of applications work really well. You just have to decide if the effort is worth the cost or, as they say, if the juice is worth the squeeze. While there is a lot of sales hype around tone detection, in my opinion, it works decently in certain circumstances. Tone detection requires training. Sometimes certain accents can be misinterpreted as escalations, and in specific businesses, words or phrases can be marked as escalations when they really aren’t. In finance, I’ve seen AI react to terms like EBITA, Dilution, Gross Margin, Hostile takeover, etc. So again, we come back to AI management. Yes, you can train AI not to react to these terms, but it requires constant attention and management. Your worst-case scenario is your staff starts disregarding AI suggestions or escalations because of all the false positives or misinformation it provides.

In conclusion, while the integration of AI in telecommunications holds immense potential, it is essential to approach its implementation with careful planning and ongoing management. The evolution from early AI applications, such as call center automation, to sophisticated generative AI and machine learning has opened up new possibilities for enhancing customer service and operational efficiency. However, as with any powerful tool, the effectiveness of AI hinges on the quality and accuracy of the content it accesses.

By focusing on real-world use cases like Agent Assist, telecom companies can see tangible improvements in customer satisfaction and operational efficiency. Nevertheless, it is crucial to remember that AI is not a set-and-forget solution. Continuous content management, regular updates, and dedicated oversight are vital to ensuring AI systems remain reliable and effective.

FYI: If you have made it this far, I recommend checking my quick video of Astro doing his tricks HERE. Have something you want me to ask him? Submit it and I’ll send you back a video of his answer. Keep it clean, folks! Have you learned nothing of AI going rogue? My family has to live with this dude.