Agent-Based AI in Network Management: Opportunities and Security Concerns

Notice: This article was created with AI.

What’s It About?

The complexity of enterprise networks keeps growing, and more and more IT managers are turning to agent-based AI systems to cope with operations. These tools take decisions independently and in real time instead of merely issuing recommendations. A joint study by Cisco and Omdia shows that more than half of those surveyed already use such tools in production. At the same time a pronounced need for control is apparent: almost all participants demand safety mechanisms and human oversight before they trust the technology completely.

Background & Context

The study by Cisco and Omdia makes clear that a considerable proportion of IT and network managers expect a fully AI-controlled operating model in the near future. Many companies make numerous changes to their network environments every day, which makes administration harder still. The number of unresolved alerts is considerable and leads to the phenomenon known as alert fatigue: staff spend a considerable amount of time checking and processing monitoring alarms, which leads to dissatisfaction.

According to the findings of the study, companies use several tools on average that often work in isolation. This makes diagnosing cross-domain problems considerably harder. While many specialists are prepared to grant AI an autonomous role in network operations, the majority at the same time want human approvals for certain actions. According to the survey, 99 per cent of respondents demand traceable decisions and emergency mechanisms before they trust AI systems.

What Does This Mean?

Agent-based AI systems could change network operations fundamentally by taking decisions and making changes autonomously.
The high number of alerts and the multitude of tools working in isolation make the use of intelligent automation attractive.
Despite efficiency gains, the demand for transparency, safety mechanisms and human control remains central to the acceptance of the technology.
Companies face the challenge of implementing AI-supported systems in such a way that they both create relief and keep to the necessary guard rails.