The Telco GenAI lowdown: Who’s doing what and why in 2024
The Telco GenAI lowdown: Who’s doing what and why in 2024
As it’s MWC season, this is the perfect time to share some of the key insights I've gathered regarding CSPs' GenAI priorities, what they’re already doing with it, and what they say are the key factors for GenAI success.
When looking back at 2023, it wouldn’t be a stretch to say that if Taylor Swift hadn’t been named Time Magazine’s Person of the Year, then generative AI (GenAI) certainly would have (and probably should have) been.
The impact of this technology on society is far reaching, having already revolutionized so much of our personal and professional lives.
And our industry is no exception, with the promise for transformative value being just as great for service providers.
Over the past months I have met with many CSPs across North America to hear about their needs, and to explore together how Amdocs can help them benefit from the GenAI revolution.
And as it’s MWC season, this is the perfect time to share some of the key insights I gathered regarding their GenAI priorities, what they’re already doing with it , and what they say are the key factors for GenAI success.
The top 5 GenAI priorities
GenAI is still a new technology, and service providers are in the early stages of adoption. They’re still exploring. And they’re also still wary.
Ours is an industry that is heavily regulated. Data privacy and data security are two major concerns that are consistently being expressed.
The legal implications of using off-the-shelf large language models (LLMs) with public data that lacks integrity, whether inadvertently erred or intentionally manipulated, could be dire.
“I’m looking forward to continuing collaboration with our partners and customers, as we see the promise of GenAI come into deeper focus."
That’s why a top priority for CSPs is avoiding the risks involved with using public data to train generic LLMs, not only for the sake of output accuracy, but also for ensuring the protection of their customers’ sensitive data.
In terms of the use cases that are being prioritized, we’re seeing that it’s those that are slated to drive more effective customer care , as well as those that improve process efficiency.
Next in line, service providers are saying that they will want to start exploring how GenAI can improve the way they initiate and instantiate new networks as well as how it can drive automation of the DevOps infrastructure .
What CSPs are doing with GenAI right now
We are very excited to be engaging with our customers on the GenAI use cases they’ve prioritized. These include improving the bill experience , enhancingself-service, and increasing the efficiency of certain internal processes.
One such process is configure price quote(CPQ).
The CPQ process can be very complex and time consuming. But with the help of a GenAI CPQ “copilot” from Amdocs, our customers have automated the proposal creation process with great success.
This copilot is driven by our telco specific LLM which is connected to the CSP’s data, resolving the data privacy and security concern.
And the drag and drop functionality that we provide has enabled them to address the need for fast, easy, and cost effective use case creation, testing, and deployment.
The 5 success factors for CSP GenAI success
This leads us to the feedback I’ve been hearing about what it takes to be successful with GenAI. And it’s been pretty consistent.
The first success factor, as mentioned, is the ability to use the CSP’s own data in training the LLM, for ensuring data integrity, privacy, and security.
Next, a telco specific LLM versus an off-the-shelf option is critical for ensuring output accuracy. Otherwise, the model might confuse 5G with George V (King George the Fifth) or PO as post office, instead of purchase order.
Third, having a drag and drop, low code/no code environment is also important, as it eliminates the need for developers and data scientists. This addresses the need for making use case creation and deployment easy, fast, and cost efficient.
The fourth success factor is choice, as it relates to the environment in which the use cases can be developed. This is where having an LLM agnostic platform comes into play.
And the last one is flexibility in creating and testing the use case whether in a sandbox, proof of concept project, lab, or real-life deployment. This is very important for hitting the ground running, and accelerating time to value.
How Amdocs is helping
Amdocs is deeply committed to helping our customers get the most out of this compelling technology. That’s why we built amAIz, a TelcoGPT platform for LLMs and GenAI apps.
It alleviates privacy concerns by connecting to the CSP’s data. It ensures telco accuracy with LLMs that we’ve ‘telcofied’ with industry taxonomy. For ease and speed, we’ve made the environment low code/no code.
We also give our customers choice, having made the platform open and LLM agnostic, following our strategic partnerships with the leading hyperscalers, including Microsoft, NVIDIA, and AWS. And finally, for getting up and running as fast as possible, we can work in a POC, sandbox, or any other approach that aligns with the CSP’s needs.
Conclusion
These are exciting times for our industry. There’s an amazing new AI technology that has come to the fore. And I am certain that it will become an integral part of how service providers run the network, their operations, and customer engagements.
I can’t wait to see all the innovative use cases that they're going to unveil. And I’m looking forward to working together with more and more CSPs and to collaborating with more partners, as we see the promise of GenAI come to life, in all its glory.
In the meantime, if you want to see how our customers are getting GenAI done with amAIz, I invite you to meet with us MWC Hall 3, Stand 3G10, and see it in action. See you there!
