vmblog logo 2024 wht (updated)

The Accountable Network: What AI Means for NetOps in 2027

Share: 

Rekha Shenoy | Published: November 19, 2026
vmblog predictions 2027

By Rekha Shenoy, CEO, BackBox

The shift in network operations is already underway, and AI is moving beyond copilots and recommendations toward systems that can diagnose issues, plan responses, execute approved actions, and even support rollbacks. Yet many people in the operations chair aren’t ready to hand over the keys, and their caution is reasonable. For NetOps leaders heading into 2027, the most important question isn’t whether to use AI. It’s how to use it while keeping humans accountable.

Prediction 1: Trust will be built offline, not in the traffic path.
The teams that succeed will start with capabilities that don’t touch live traffic. Reading thousands of pages of vendor advisories, determining which vulnerabilities apply to a given environment, and flagging which workarounds are already in place are the kinds of “smart intern” tasks that can save hundreds of hours without granting AI administrative access. Pace matters here, too. The patch apocalypse is real. Attackers have AI, too. Vendors are urging faster upgrades, and the volume of patching work has outgrown what human teams can absorb. One customer calculated that doing patching properly would take 23 people, and no organization is going to hire that many for the job.

Prediction 2: AI will draft the work, and humans will approve it.
The next step is AI that builds automation but doesn’t run it. It generates a remediation task, presents it visually so an engineer can understand it, and waits for approval. The engineer stays accountable and chooses what to run, while days or weeks of effort shrink to minutes. This also helps address a persistent skills gap. Engineers shouldn’t have to be experts in every vendor’s device behavior or strong coders. Automations tailored to specific platforms take that burden off them.

Prediction 3: Autonomy will be earned over time, with a named human behind it.
After a year or eighteen months of approving recommendations, many operators will find that clicking “OK” has become a rubber stamp. At that point, some will authorize automatic remediation, with a real person, whether a frontline engineer or a director, formally owning the decision. Even then, safeguards will remain: back up before every change, test afterward, back up again, and keep clear off-ramps so ambiguous or unfamiliar cases go to a person. A system that is right 99.5% of the time still needs a plan for the other 0.5%.

Prediction 4: Data quality and provenance will become a competitive advantage.
AI built on messy or unverified sources produces garbage with confidence. Operators should ask vendors how their models are built, what data they learn from, and whether the underlying automations are clean on each platform. Provenance, meaning what changed, why, who approved it, and where the ticket lives, becomes more valuable as automation scales. It’s also what lets an engineer quickly regain control when something goes wrong after months of trusting the system.

Prediction 5: The NetOps role will grow, not shrink.
Predictions that there will be fewer network operators are likely wrong. The more probable future is a broader remit and a more interesting job. Engineers will think more like architects or managers of agents, focusing on end-to-end behavior and business context, such as a hospital network where downtime in the surgical suite is unacceptable. They will still need deep expertise because verifying that an agent did the right thing requires knowing how to do it yourself. The driving analogy fits: lane assist and adaptive cruise control reduce fatigue, but the driver stays alert and retains the skills.

Prediction 6: Integration and budgets will reshape how teams buy.
AI will connect tools that currently sit in silos. Observability, backup, ticketing, and CMDB platforms will feed agents that can trace a 2 AM outage to the exact configuration change in minutes. Budgets will shift as well. Infrastructure teams that lost funding to cybersecurity may find AI budgets available for tools with legitimate business outcomes. Some organizations will build their own automation; others will buy it, and both choices can be sound.

The teams that thrive in 2027 won’t be those that resisted AI or adopted it recklessly. They will treat it as one more tool in the toolbox, hold vendors to a higher standard of accountability, and remain firmly in control of its use.

##

ABOUT THE AUTHOR

rekha shenoy

Rekha Shenoy is the CEO of BackBox and a highly recognized thought leader in network cyber resilience. With 25+ years of B2B tech experience, she previously drove the $710 million sale of Tripwire and led the rebranding of a $100 million product line at BMC Software.