HPE Aruba boasts that when network problems come along, its AI will whip them into shape

NetAdmins may be mere years away from devolving into babysitters for bots

Not all the autonomous agentic AI that HPE announced at its annual Discover conference this week is live and ready for customers, but don't tell that to the Aruba networking group – whose enthusiasm outpaces its parent company’s, at least in terms of talking points.

HPE went all in on agentic AI at this year's event – including, as we reported yesterday, the addition of agentic AI mesh technology and a conversational networking copilot (a.k.a., a chatbot) in Aruba Networking Central.

The new tech places task–specific agents under the authority of a multi–agent orchestrator (MAO) that can aggregate data supplied to it by its sub agents. The MAO communicates with humans using the newly announced networking copilot, which admins can ask to gather data, perform root-cause analysis, make remediation suggestions – or take action on its own.

These features, Aruba SVP and GM Phil Mottram said in a Tuesday briefing, are like "having an extra set of free network administrators sitting beside you." He predicted those virtual admins "are going to be very helpful to our customers."

As one example of what Aruba's new AI features can do, Mottram used the example of the 802.11r specification, which aims to improve the speed at which devices connect to Wi–Fi access points as users move beyond the range of one radio and come closer to another. The specification is known to work quite well with Apple devices, but Android support has been a consistent mess.

Mottram explained that an autonomous networking agent could see a degradation in service, determine that there were a large number of Android devices on a particular Wi–Fi access point, and turn off 802.11r to address the problem – and even switch it back on when fewer Android devices attempt connections.

"That's the sort of thing that it would take a while for a network admin to understand and then change," Mottram said. "An intelligent platform that's looking at user experience across the network can find that problem quite quickly."

Mottram told us the 802.11r repair scenario is available today.

According to Seelan Manavalan, VP of product for Aruba, the new version of Aruba Central includes around 200 pre–packaged, AI–powered autonomous capabilities. Automated agentic AI in the package can also do things like optimize energy usage and repair simple issues – basically anything with a straightforward root cause analysis and response is fair game for the new Aruba agentic mesh.

That said, there's some question about how these kinds of pre–packaged scenarios constitute "AI" versus simple old–fashioned automation, but perhaps baby steps are appropriate, as many customers have chosen not to let the MAO take complete control of things within its scope quite yet.

"Most customers are still trying it out and getting comfortable with letting an agentic AI act autonomously," Manavalan told us.

Mottram, likewise, said that customers are still "on a journey to trust the technology," but it's ready when they are – at least for the given use cases in which Aruba's agentic mesh can operate autonomously.

Those use cases will have to expand before network engineers can get hands off and become babysitters for agents, but Manavalan doesn't think that transition is a distant prospect.

"We will have autonomous networks in the next few years," the Aruba VP predicted. Hopefully this won't mean datacenter network administrators will need to dust off those resumes, but the AI shift is definitely starting to impact the IT jobs market. ®