The promise of artificial intelligence in enterprise environments has long oscillated between visionary ambition and pragmatic hesitation. Today, that tension is reaching a critical inflection point within the telecommunications and IT sectors. Agentic AI—systems capable of not just analyzing data and offering recommendations, but proactively taking direct action—holds the transformative potential to make modern networks smarter, predict and resolve impairments before they impact end users, and significantly alleviate the daily strain on IT professionals.
However, as companies increasingly hand over the operational keys of complex digital infrastructure to autonomous digital teammates, a profound underlying question remains: Exactly how much responsibility should organizations cede to algorithms, and how is that trust earned, measured, and maintained?
According to a comprehensive new study from Omdia, backed by data and analysis from networking giant Cisco, establishing and maintaining trust in artificial intelligence is a critical yet moving target. As agentic AI sees wider integration into everyday network operations (NetOps), the parameters surrounding accountability, visibility, and control are shifting rapidly.
The findings suggest that debating whether agentic AI should be deployed is already an outdated exercise. The momentum of adoption has far outpaced preliminary industry discussions. The Omdia study, which surveyed roughly 1,000 IT and network operations professionals working within organizations boasting more than 500 employees, reveals that nearly three-quarters of enterprises have already deployed AI for network operations. Most notably, 51% of those respondents reported that the agentic AI systems currently active in their environments are already operating in an active capacity—making decisions and executing tasks rather than merely serving in an advisory role.
Even amidst lingering industrywide debates regarding safety guardrails and system transparency, four out of five survey respondents expressed comfort with granting AI a "high or fully autonomous role" in managing network operations. This cohort includes 24% of professionals who stated they are completely comfortable with fully autonomous configurations that operate entirely independently of human oversight. Looking ahead, the operational shift is projected to accelerate even further: 84% of surveyed professionals anticipate that their respective companies will implement an "AI-led" operating model within the next 12 months.
IT Can’t Hire Its Way Out of the Problem
The rapid pivot toward autonomous network management is not merely driven by a desire for technological modernization; it is largely being propelled by an acute operational necessity. Modern enterprise networks have grown exponentially in scale, distribution, and architectural complexity. Simultaneously, the volume of telemetry data and automated monitoring alerts generated by these environments has surged past the cognitive processing limits of human engineering teams.
The Omdia study highlights a striking operational imbalance: the average enterprise organization generates approximately 4,100 monitoring alerts and events every single day. Of that total, more than half—roughly 51%—are directly related to network performance and infrastructure health.
When translated into human workload, the math reveals an unsustainable reality. Industry estimations from Omdia and Cisco indicate that a typical, highly skilled network practitioner is realistically capable of thoroughly reviewing, investigating, and resolving roughly 21 network alerts per day. To clear the daily backlog of a standard enterprise environment manually, an organization would need to maintain a dedicated team of approximately 100 network specialists. For the vast majority of companies, maintaining an operations team of that scale is financially and logistically impossible.
This widening gap between operational demand and available human capital has created an environment where automation is no longer optional. Joe Vaccaro, Senior Vice President of Network Platform and Assurance at Cisco, emphasized the severity of this workforce crunch during discussions surrounding the new data.
"IT is managing more with fewer people than ever before," Vaccaro noted. "And network connections have become more complicated than ever before. IT just can’t hire their way out of this anymore."
Faced with this labor shortage and escalating complexity, enterprises are turning to agentic AI not to replace human ingenuity, but to act as a crucial force multiplier that can triage, diagnose, and remediate routine issues before they cascade into catastrophic outages.
AI Must Show Its Work
The widespread integration of autonomous agents into critical infrastructure does not happen in a vacuum. This operational evolution unfolds against a backdrop of public headlines and legitimate corporate anxieties regarding autonomous systems exhibiting unexpected behaviors—such as instances of AI agents acting unpredictably or "going rogue" before human supervisors realize a deviation has occurred.
Consequently, establishing robust security, comprehensive governance frameworks, and deep-seated trust has emerged as a paramount priority for enterprise technology leaders. Bringing an autonomous software agent into an enterprise network environment requires the same deliberate vetting process traditionally reserved for human personnel.
"You need a level of trust as you bring [agents] into the fold as you would with a new junior network operations engineer," Vaccaro explained.
Even as organizations embrace high levels of automation, human oversight remains an indispensable component of the operational architecture. Rather than relinquishing total control unconditionally, enterprises are demanding rigorous accountability mechanisms. A full 99% of survey respondents indicated that they require specific guardrails and verification protocols before they are willing to trust an AI system to take independent action. Furthermore, 69% stated that they actively demand real-time tracking and logging of all agentic actions executed by the software.
This emphasis on auditability points to a central requirement for the next generation of enterprise AI tools. Systems cannot simply execute changes behind a black box; they must be transparent in their reasoning and execution.
"The core thing we see through the survey […] is the importance of [AI agents] showing their work," Vaccaro said.
Beyond mitigating risk and ensuring uptime, the adoption of agentic AI is rapidly redefining corporate productivity benchmarks. The Omdia study uncovered a strong correlation between high levels of AI integration and overall organizational efficiency. Companies that have achieved the highest tiers of AI adoption within their network operations demonstrated productivity levels nearly 30% higher per employee compared to their less mature industry peers.
This performance differential indicates that AI maturity is shifting away from being a supplementary operational improvement and is rapidly becoming a decisive competitive advantage in the modern digital economy. However, unlocking these productivity gains requires more than just purchasing advanced software licenses; it demands structural alignment across the enterprise.
"The biggest barrier might not be technology in the first place," Vaccaro noted, pointing to the human and structural elements of digital transformation. "It might be organization readiness."
