The networking world is undergoing a critical transformation. As environments grow more complex, talent shortages deepen and IT expectations rise, the industry is broadening its search for better ways to manage and secure the backbone of digital businesses. Enter bespoke AI solutions. This new crop of tools is designed to deliver precision, accuracy and accountability.

Meter Command is one such tool. At its core, it delivers a purpose-built AI model designed specifically for networking, according to Anil Varanasi (pictured), chief executive officer of Meter Inc.

“I’m curious how much of the audience, either listening to this or watching this video, would be comfortable if we took a chat [large language model] and put it in one of their cars and said, ‘Drive it,’” he said. “Not that many people would be comfortable getting behind the wheel of a Tesla or a Waymo if it were driven by a language model that’s not built for it. Similarly, none of us in networking should trust models that are not built for networking to take action on a network, drive a network to the outcomes that are needed and have precision and accuracy for the actions that need to be taken.”

Varanasi spoke with theCUBE’s Bob Laliberte at The Networking for AI Summit event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed Meter Command as a turning point, ushering in the era of bespoke AI tools, collaborative workflows and safeguards for trust to redefine how networks are run. (* Disclosure below.)

Bespoke AI moves networking beyond command-line interfaces and dashboards

Over the decades, network management has evolved from CLI to web dashboards. But as networks grow in size, complexity and user demands, both approaches are reaching their limits. Command represents the next iteration: A natural language interface that allows engineers to query, configure and even build software instantly, according to Varanasi.

“The last thing that Command can do is actually just write software entirely on the fly,” he said. “We believe software should be soft and that it should be delivered to end customers the way they want it at the moment they want it. Our models are trained on not just our hardware architecture, our data pipelines and software, but also how we design and write software as well.”

One of the biggest concerns with AI-driven automation is trust. A single misconfiguration in a production network can have far-reaching consequences. To address this issue, Command includes built-in confirmation and validation workflows. Before executing any action, it reviews the request, presents a summary of potential outcomes and requires user confirmation, Varanasi added. This “human-in-the-loop” design helps ensure that engineers remain in control while benefiting from AI-driven speed and efficiency.

“If you think about how humans know how to ride a bike, you start with training wheels, then a parent or an older brother or older sister is holding your bike,” he said. “Then you finally are like, ‘Okay. I can ride on my own.’ I think in any system that we all use — any system we all trust — there are levels to it. And as you gain confidence, you build trust.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of The Networking for AI Summit event:

(* Disclosure: TheCUBE is a paid media partner for The Networking for AI Summit event. Neither Meter Inc., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE