10 examples of how telcos are using AI (BESIDES customer support!)
Last week I had a great meeting with one of the cofounders of Anthropic. At one point our conversation turned to all the amazing things telcos can do with AI. While we naturally started the discussion talking about co-pilots being deployed in customer support call centers (DUUUUUH), we both agreed that generational AI (GenAI) can do SO MUCH MORE than that.
Perhaps the biggest and boldest moves are being led by South Korea’s SK Telecom, which has professed to becoming an “AI-first” organization (which is a much better idea, IMHO, than a “telco to techco” transformation). So, what’s next? What other ways can a telco use AI to improve? Here are 10 real-world examples of how trailblazing operators are using AI to deliver real value to their organizations.
1. Build a telco-specific LLM: Global Telco AI Alliance
Claude by Anthropic and ChatGPT by OpenAI are general-purpose large language models (LLMs) known in the AI industry as “foundation models.” They’re able to handle, or be adapted to, a wide range of tasks. But like someone with general knowledge, they’ll be out of their league when it comes to anything that requires domain-specific, in-depth knowledge—like what business users with deep telco expertise need. For that, you need task- or industry-specific LLMs. Does telco need one? A new consortium says yes.
The Global Telco AI Alliance—comprised of SK Telecom, Singtel, Deutsche Telekom, e&, and SoftBank—is going for the big AI kahuna. The joint venture aims to build a telco-specific LLM, with equal investment from each member. The goal is to provide better service to their combined 1.3 billion subscribers across 50 countries in a number of languages: Arabic, Bahasa, English, Korean, and German. Two questions: when it’s ready, will the Alliance keep it to themselves, or let the rest of the telco world use it? And will the effort and expense be worth it in the end, or will most telcos have gotten what they need with generic LLMs and a Retrieval Augmented Generation (RAG) approach? (This episode of the podcast Will 5G change the world attempts to answer the second question.) Only time will tell.
2. Code generation: Totogi
The Anthropic co-founder and I agree that the REAL power of GenAI is code generation, and telco is lacking in this area. It’s time to say hello to Totogi’s BSS Magic.
A slew of STL customers (acquired by Skyvera in 2023) are all using BSS Magic to complete CRs in record time and for a lot less money. This AI-powered tool uses TM Forum’s Open Digital Architecture (ODA) and Open APIs to help telcos customize their existing multi-vendor business support systems (BSS) quickly and easily. Business users use natural language to describe the changes they want in their BSS for things like updating screens, adding new functionality in their system, or modifying the workflow. BSS Magic generates the code and adds it to their BSS without needing technical folks to write code themselves. Check out this video to learn more.
3. Network optimization: Three UK, AT&T, and Ooredoo
Telcos generate and store a truckload of data, and have struggled in the past with being able to crunch it to find the valuable, actionable insights that can help improve service quality and subscriber experiences. But today we have AI, and it has the chops to do the job.
For example, Three UK handles 29% of the country’s mobile data traffic and uses Azure Operator Insights from Microsoft to quickly analyze data, identify areas for improvement, and make informed decisions to improve service quality. The AI-powered approach allows it to deliver responsive, high-speed, low-latency services, ensuring an optimal experience for the growing number of gamers, streamers, and social media denizens asking more and more of its network.
And Three UK isn’t the only one striking network gold with the savvy use of AI. AT&T is using AI, machine learning (ML), and predictive analytics to design, build, and maintain networks, including making decisions about acquiring spectrum and locations for cell sites. AI/ML also helps the US telco improve forecasting and capacity planning, validate new equipment, meet customers’ network capacity demands, detect network issues in real-time—and fix them. AI-based automations also help AT&T use network resources more efficiently, reducing its carbon footprint. As an early AI adopter, AT&T hasn’t been shy about the billions it has saved, and will continue to recoup as it reshapes itself for the future.
Ooredoo is also using AI for network optimization, having deployed a new solution from Ericsson on Azure. It uses digital twin technology and advanced AI techniques to analyze the radio access network and provide optimization recommendations as well as resolve specific performance issues. The result: a better experience for subscribers and lower operating costs for Ooredoo. Smoother calls, faster downloads, and fewer frustrating “network busy” messages: what’s not to love?
4. Expanding fixed network access: Vodafone Germany
Increasing network availability and reach can be complicated and expensive, with a lot of variables to consider from geography, to capacity, to technology. Luckily, AI is perfectly suited to analyze a lot of variables!
Vodafone Germany uses a proprietary AI algorithm to plan the expansion of its fixed broadband network. The algorithm creates rough network expansion plans for all of Germany in a few minutes, factoring in cost, time, and efficiency to deliver an optimal cost-benefit ratio. Then a team of human experts takes over from there. Compared to the old approach, AI has made the process five times faster.
5. Cloud management: Globe Telecom
AI is the cloud whisperer, making complex cloud management a breeze! It’s like having a consultant making sure everything runs smoothly.
Philippines-based Globe Telecom saw the writing on the wall regarding moving workloads to the public cloud years ago, and acquired cloud management platform Cascadeo back in 2019. Last year, the Cascadeo business unit of Globe launched an AI-powered version of its platform, Cascadeo AI v3.0. Multiple GenAI tools—OpenAI GPT-4, Amazon Web Services (AWS) SageMaker, AWS DevOps Guru, and proprietary tech from Cascadeo—have helped to enhance, simplify, and automate cloud management, enabling Globe’s enterprise customers to monitor, manage, and optimize cloud deployments in real time. If you’re running cloud workloads for your enterprise customers, you should think about providing an AI-powered management platform for your customers, too.
6. Employee support: AT&T and TELUS
Internal use cases are a great way to test out AI. It gets your team familiar with work with AI on a daily basis, coupled with lowered expectations on quality. Creating AI-powered tools that help employees by answering company-related questions and performing everyday tasks is a fantastic way to experiment and rapidly iterate with AI.
AT&T built an internal, AI-powered tool called “Ask AT&T” via access to an early version of ChatGPT, thanks to its partnership with Microsoft Azure. The tool now helps over 68,000 employees with writing code, customer service, summarizing meetings, analyzing network data, and other tasks. The deployment has been so successful that AT&T is working on training the tool on more data so it can handle even more and different kinds of queries. The company is also preparing for the NEXT phase of the revolution with autonomous assistants that are helping with fraud alerts, software development, and network optimization.
Once your people are comfortable using GenAI at work, you can extend that experience to customers. Note that chief data officer Andy Markus mentions in this blog that once open-source or cheaper large language models (LLMs) are able to do the job, AT&T would be happy to swap one in for ChatGPT to save some coin.
Another North American operator is also using AI for employee support. Canadian-telco TELUS launched Fuel iX, an AI-driven platform that taps into more than 100 LLMs to help both internal teams and enterprise customers use and monitor AI across different systems and clouds. Since TELUS started using the platform to boost its ability to predict network issues, forecasting accuracy has skyrocketed from 69% to 89%. What a great way to help field technicians work smarter and more efficiently. Fuel iX is also contributing to cost savings, reducing network operating costs by almost $15 million dollars.
7. Employee recruitment: Deutsche Telekom (DT) and Omantel
Remember when chatbots were the new-new thing? Little did we know they were the GenAI bellwether, demonstrating how much work they could save us in the area of recruiting.
DT was a chatbot pioneer, launching one in 2016 to field job seekers’ questions about its hub:raum startup accelerator. Built on Facebook Messenger, the bot is happy to engage after hours, when human recruiters aren’t available. It can talk to multiple people at once, never gets tired of answering the same questions over and over, and never stammers when asked direct questions about pay. Candidates liked the experience and applicant quality increased! Now, other DT subsidiaries are incorporating recruiting chatbots to answer questions as well as let people search, browse, filter, and select jobs from within the bot.
And Omantel used Elevatus’ AI-powered video interviewing technology to screen more than 5,200 applicants for its Gen Z program in just two days—that’s serious efficiency. This is the kind of innovation that’s transforming telco recruitment and helping companies find top talent at lightning speed.
8. Subscriber identification: Telefónica
When I think of voice detection, I think of spy movies where characters use voice modifiers to disguise their identity. But you can use AI and make it way cooler and more creative, like this example.
To identify senior citizens calling in for support and give them priority service, global telco Telefónica uses Nuance Gatekeeper, a security solution that authenticates people through voice. When the COVID-19 pandemic forced seniors to rely on their phones for everyday interactions like doctor and pharmacy visits, Telefónica recognized the urgency of assisting this at-risk segment of the population and moved them ahead in customer service queues. You’d think asking callers if they were over 65 would be easy enough, but it didn’t work. Everyone figured out the trick to getting ahead in the line and it led to a lot of dishonest replies from wannabe line-jumpers (BOO!). So, the operator engaged Nuance to help validate the real seniors using voice characteristics, leaving the fakers at the back queue. Brilliant!
9. Revenue leakage: Lebara
AI is the Sherlock Holmes of the telecom world, sniffing out people trying to game the system in big and small ways at scale, and instantly. Here’s a more traditional approach to fraud detection.
Leading UK mobile virtual network operator (MVNO) Lebara uses the SageMaker tool from AWS to identify and block all kinds of fraud, including SIM boxing, revenue share fraud, and good-old wangiri fraud. The process uses call detail records to build, train, and deploy machine learning models. Here’s a presentation from an AWS Summit event where Lars Hoogweg, then-CTO of Lebara, explains. (His bit starts at 11:35.) Even with limited knowledge of ML, Lebara was able to use SageMaker to achieve previously undetected fraud results within a few days of its ML model going live. Imagine how well it’s performing now, after years of fine-tuning?
10. Plan personalization: Mobile X and Totogi
Forget one-size-fits-all plans! AI can create tailored plans custom-fit each customer based on any number of characteristics.
For example, do you know how much data you need? Startup MVNO MobileX is going after the US market with the belief that most mobile subscribers don’t. Instead of pushing unlimited plans, the company offers an AI-based tool that analyzes each subscriber’s usage and creates personalized plans for them. Hear all about it from CEO Peter Adderton in this Telco in 20 episode.
And don’t forget Totogi’s work to optimize revenue at the individual subscriber level with micro-targeted plans and offers in our PlanAI product. We use ML and AI to generate cohorts of subscribers to increase revenue based on charging data. A few of our customers are already piloting the technology with the goal of lifting revenue by 10%.
What’s next?
This is by no means a comprehensive list, and no doubt there are new AI use cases in development right now. There will be more telcos creating their own AI-based solutions and leveraging new products and services that come to market in the next few months and years. Working on a cool and innovative AI project yourself? Tell me about it!
Recent Posts
Get my FREE insider newsletter, delivered every two weeks, with curated content to help telco execs across the globe move to the public cloud.
"*" indicates required fields
Listen
Microsoft Azure is helping telcos manage their data for AI workloads and prepare for an AI revolution.
Watch
DR opened MWC’s second annual MVNO Summit with an epic talk about the AI-powered revolution in telco business support systems (BSS).
Read
Think AI is too expensive for your org? Think again. When implemented strategically, AI can pay for itself.