BT’s network platform strategy is rooted in AI

  • AI and LLMs will play a role in BT transforming its network into a programmable platform 
  • AI is enhancing BT’s automation and enabling real-time services for customers, BT's Alex Healing said in an exclusive interview
  • BT will continue to keep humans involved in critical network and cybersecurity decisions

Artificial intelligence (AI) isn’t a new player at British Telecom (BT) — the company has been leveraging AI solutions for decades. But what’s happening now, with advanced technologies like generative AI (GenAI) and large language models (LLMs), has the potential to reshape BT’s network into a programmable platform.

“BT is moving away from dictating the network services that it offers as a company, and moving toward more of a network platform,” said Alex Healing, BT distinguished engineer and research principal. As its networks evolve, the operator “sees AI being a big part of that evolution.”

The cornerstone of BT’s network transformation strategy is programmability, Healing told Fierce Network — integrating software, open APIs, advanced telemetry and automation into the heart of the network. Ultimately, the goal is to make the network more flexible and on-demand, so that customers can access services in real-time as needed.

Moving toward a network platform can “unleash product innovation on top of the network, exposing network capabilities, network services, for others to develop on top of, as well as developing our own capabilities,” Healing said.

This week BT announced it is officially turning on its "Global Fabric" network-as-a-service platform, with services to launch in early 2025 after successful live testing. The platform aims to make it faster and easier for businesses to connect securely to apps, digital services and AI tools across multiple clouds.

The 'softwareification' of the network

BT’s shift toward a software-driven, cloud-centric network is reflective of broader trends across the telecommunications industry.

“AI is a fast-moving field, as is the network. So obviously, we're seeing the softwareification of the network, with trends toward the cloud and software and network space,” Healing said. “Likewise, of course, AI is developing as well.”

BT has been using AI tools for decades (since the 80s, according to Healing) to optimize planning and network operations. But now, advancements in GenAI and LLMs are helping BT build a network that’s more dynamic.

More recently, the company’s researchers are “looking at things like assurance of the network and the services that run over that network,” Healing said, adding that BT researchers are also looking to plug more AI capabilities into the network's security functions. 

Humans at the helm

Despite its growing reliance on AI, BT is taking a balanced approach to integration.

While network autonomy is certainly a long-term goal, the operator is not in a rush to fully automate the network, noted Healing. Like many of its peers, the company will keep a "human-in-the-loop" approach, ensuring human oversight remains central to critical decision-making processes.

“We're working toward that programmable, that autonomous network,” Healing said, “ but I think certainly the way we see things is that a human in the loop is going to be the case for a lot of that operation.”

BT is already applying this human-centered AI approach in cybersecurity. Healing highlighted the company’s use of AI to build knowledge graphs of cyber threats, but with human analysts responsible for final decisions. Similarly, LLMs are being used to process vast amounts of unstructured data — like network logs — while humans retain control over critical operational choices.

As BT moves toward its goal of a fully programmable, platform-based network, it will continue to evaluate where automation makes the most sense — and where human input remains necessary. “We want to leverage AI to help make things better. We don’t want unwanted effects occurring,” Healing concluded.

“I think that's part of that process of implementing an autonomous network. Where is it that you can most benefit from the automation? And how do you achieve that automation through what techniques — is it through simpler techniques, statistical techniques, process automation techniques, or is it actually the more complex, maybe Gen AI, techniques that really are required?”

Editor's Note 10/1/2024 2:30 p.m. ET: This article was updated with information about BT's "Global Fabric" network-as-a-service platform announcement.