Using GenAI to enhance network automation

Every once in a while, life confronts us with the inevitable, and generative artificial intelligence (GenAI) is triggering one of those moments now. It’s taking our networks into the next era, where an AI assistant helps us oversee their growing complexity so we can offer our customers a richer and better experience.

GenAI gives us a lot to talk about

The GenAI industry has become a giant, with hundreds of billions of dollars spent by the biggest public cloud players, as well as hundreds of GenAI start-up companies. Hundreds of large language models are in a race to become bigger and better, with the latest models well beyond 1 trillion parameters and requiring well over 200 GB of memory to perform.

On the other side of the spectrum, there is an ongoing “small is beautiful” movement that’s allowing language models to become highly specialized in specific domains, or to be able to run even on mobile handsets. The use cases these language models aim to support are endless, and the number of people that could potentially benefit from GenAI is, well, the entire population of the planet. In short, GenAI is about a lot of everything, which raises the question: Where do we start?

Let’s go back to the beginning

We have all had that feeling where we long to have an assistant by our side, ready to help us out with a quick start, some guidance for selecting the best alternative or a thumbs up when a major milestone has been achieved.

Let’s take this line of thinking back to where network automation started: a few people exchanging ideas about how to talk to a few routers, together forming a small network, and listening to those routers when they need to get a message across. Read this sentence again: exchange ideas, talk to, listen to. This does sound like language, which could take the form of English, code and alarms, respectively. Out of all the stakeholders involved with network automation, we could start with only a few developers and a few routers, and use GenAI to connect different forms of languages.

Leveraging GenAI to deliver artifacts

Network automation is a hot topic in the networking industry. The high expectations for network automation are driven by the proliferation of principles such as network models, network APIs and SDN controllers. This promises to separate the network automation platform from the network services, which are delivered as software artifacts on top of this platform. This approach is not much different from the current practice in the software and cloud industries, and it’s worth mentioning that there are currently quite a few large language models that specialize in providing assistance with generating code.

Network service developers need to communicate, and that requires a large language model that isn’t loaded with vocabulary expressing emotions, feelings or beauty of any sort. It also requires a small language model with knowledge of a network API that is attached to software code and the commentary included in that code.

Of course, these language models need to be fine-tuned with our networking vocabulary, our network automation platform documentation and our software code, with its commentary from all artifacts that have already been generated and tested. Adding the test descriptions and results will be valuable as well.

With a GenAI framework that ties all this knowledge together, we are prepared to add true artifact assistance to our network automation platform.

NSP brings together all stakeholders of your network services

Nokia Network Services Platform (NSP) is the network automation platform of choice for the world’s most advanced communications service providers because it allows them to deliver advanced network service intents tuned to their customers’ specific needs. NSP’s stakeholder group extends from the Nokia teams that develop and test the automation platform and the artifacts that will be added on top, over to the Nokia engineers that tailor the artifacts to each customer’s network to provide the best experience, and all the way to the customer’s design, planning and operation teams.

The addition of GenAI capabilities allows NSP to provide all stakeholders across the value chain with the right assistance, be it artifact, knowledge, troubleshooting and any other type of assistance that helps you bring new network services to market more quickly and efficiently.

Our vision is that the GenAI-enabled NSP will act as an assistant for stakeholders in your networking departments and business-to-business and business-to-consumer markets. This vision culminates in a network that serves the success of your business in a very direct, reliable and differentiating way.

Find out how we are integrating GenAI more extensively into NSP for automating network operations: Nokia IP networks and AI