AI agents tackle network scaling issues

Network teams are struggling to keep up with the demands of AI and automation as companies push for instant issue resolution and rapid application deployment across internal and cloud environments. Developers already use advanced automation, but network professionals lag behind, raising the question: can AI agents help close the gap?

These agents aim to extend automation beyond observability and AIOps, handling monitoring, alerting, incident response, and even security and compliance. Over time, they could autonomously scale and manage network resources to optimize workloads.

To function, AI agents need consistent business rules from network, security, and compliance teams. They use machine learning to improve performance, but deployment remains aspirational for most organizations.

Why adoption remains limited

Uniform rules across data centers, edge, and cloud networks are hard to maintain. Many companies also face integration challenges and poor coordination between security, compliance, and network teams, leaving AI agents with incomplete guidance.

Staff learning curves add another hurdle. While teams have moved from basic monitoring to observability, engagement with AIOps—seen as a stepping stone to AI agents—is still limited. Vendors do offer migration paths, but adoption isn’t widespread.

A test case in action

A November trial by Nanites demonstrated an AI agent resolving a simulated Cisco IS-IS network outage in three minutes—work that typically takes a human engineer over 30. The system autonomously analyzed alerts, identified root causes, and executed fixes in seconds, though it required human approval.

The trial was conducted in a tightly controlled environment, not the hybrid, multi-cloud setups most enterprises use. This highlights a key limitation: even in ideal conditions, full autonomy isn’t trusted yet.

Pressure to scale

In February 2026, Neraj Kumar, director of solutions engineering at SolarWinds, cited IDC research showing 59% of organizations investing in AIOps to automate monitoring. Yet 75% remain focused on “keeping the lights on,” hindered by tool sprawl and data overload.

“No CIO walks in on Monday and says, ‘My environment is simpler than it was last year,'” Kumar said. Hybrid and multi-cloud adoption has increased flexibility but also complexity, with more integration points and telemetry streams.

Preparing for AI agents

Organizations can start by identifying which operations they’d most want to automate. Setting clear priorities helps focus strategy on meaningful outcomes.

Next, teams should define a roadmap, accounting for real-world friction. The Nanites trial proved AI agents can perform under controlled conditions, but human oversight remains critical. Vendors should be chosen based on long-term support and product investment.

Finally, testing AI agents in controlled environments first—before scaling to broader, more chaotic networks—can help iron out issues. The technology is still early, but the groundwork can be laid now.