AI has firmly moved beyond experimentation in enterprise IT. In 2026, it is embedded in day-to-day operations, including the network, which is evolving to meet the speed, scale, and adaptability that AI-driven systems demand – handling continuous data flows, adapting to shifting workloads, and supporting systems to run faster, smarter, and more reliably.
Head of AI Engineering & EMEA CTO at Extreme Networks.
Despite these advances, full autonomy remains rare. While 89% of leaders would trust AI agents to take specific, narrow network actions without human oversight, only 10% would allow fully autonomous decision-making.
Article continues below
This sums up where organizations are right now: they are embracing AI-powered automation, but they’re not ready to relinquish control entirely. Instead, they’re landing somewhere in the middle, in a phase of guided autonomy.
Why trust hasn’t yet translated into action
The hesitation is not about whether AI works, its value is already being demonstrated. AI is now used in core networking tasks such as performance monitoring, anomaly detection, capacity planning and troubleshooting, where speed and responsiveness are essential for most operations.
These are also the areas where organizations are reporting the most immediate benefits.
At the same time, the network plays a uniquely critical role supporting AI and agentic workloads across the enterprise. It underpins customer experience, supply chains, financial transactions and a wide range of business operations. When something goes wrong, the impact is immediate and visible.
Because of this, organizations remain cautious about letting AI act without oversight, and human-in-the-loop rightly remains the default for most operations. Concerns around accountability, transparency, reliability and governance continue to be key barriers, even as confidence and trust in the technology grows.
Until organizations fully understand how AI makes decisions, and how those decisions can be controlled, many are not yet ready to remove humans from the loop entirely. Even so, the benefits in efficiency, speed, and AI supported analytics and decision-making are already substantial.
The rise of guided autonomy
Rather than choosing between full manual control and complete automation, most organizations are adopting guided autonomy.
In this model, AI handles routine or lower-risk tasks such as performance optimization, traffic routing or diagnostics, while humans retain oversight for critical decisions, policy changes, and security-sensitive actions.
Clear boundaries, governance, and visibility are built in to reduce risk while enabling AI to act efficiently. Guided autonomy lets organizations leverage AI’s speed and scalability without sacrificing oversight, bridging the gap between trust and action and maintaining a level of control aligned with their risk tolerance.
Making guided autonomy work
Guided autonomy offers a practical way for organizations to benefit from AI tools while maintaining control, but it requires thoughtful design.
Modern networks generate massive volumes of data, support distributed applications, and respond to real-time user and device demands. Without clear boundaries, requiring human approval for every decision can slow processes and limit AI’s efficiency.
This careful oversight is what makes guided autonomy effective. By combining AI’s speed with human judgment, organizations can automate routine tasks, such as performance optimization, traffic routing, and diagnostics, while keeping critical decisions, like policy changes or security-sensitive actions, under human supervision.
The focus is on where and how oversight is applied, not on restricting AI. Done well, this approach allows businesses to accelerate operations, improve responsiveness, and build confidence in AI-driven decision-making, all while maintaining the right level of control.
Building accountability into autonomous systems
As AI takes on a greater role in networking, accountability becomes just as important as capability. AI systems can no longer be treated as simple tools; they behave more like participants in the network, which means their actions must be governed, monitored and auditable. Ultimately, however, humans retain responsibility.
This also requires greater visibility, transparency and explainability. Organizations need to understand not only what is happening on the network, but why. Without that context, trust is difficult to build, and even harder to maintain.
Encouragingly, IT leaders now view AI as a way to reduce risk, rather than introduce it. By detecting anomalies earlier, enforcing consistent policies, and responding faster than humans can, AI can strengthen security as well as reliability. But that only holds true when accountability is designed into the system from the start.
From supervision to orchestration
As confidence in AI grows, the role of human teams is already starting to change. Many organizations expect to reduce human involvement in routine network decisions within the next year, moving from human-in-the-loop to human-on-the-loop, reflecting a broader shift towards more autonomous, self-managing environments.
The organizations most likely to succeed won’t be the ones that move fastest, but those that move deliberately – building trust, setting boundaries and gradually increasing autonomy to strike the right balance between humans and AI.
We’ve featured the best AI website builder.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit
Leave a comment