Enterprise IT operations are entering a new phase as organizations move beyond traditional observability and AIOps platforms toward more autonomous, agent-driven operational intelligence. While early AIOps initiatives focused primarily on correlating alerts, surfacing anomalies, and improving visibility through dashboards, the next wave of operational AI is centered on action, orchestration, and decision-making across increasingly complex environments.
In a recent episode of NetworkANGLE, Shailesh Manjrekar, Chief AI and Marketing Officer at Fabrix.ai, joined theCUBE Research to discuss how enterprises are evolving from traditional AIOps toward what he described as “AgentOps,” an operational model designed to coordinate AI agents, unify operational data, and enable more autonomous IT workflows.
The discussion reflects a broader industry transition. As enterprises operationalize AI across infrastructure, applications, networking, and security, many are discovering that scaling AI requires more than deploying models or co-pilots. Organizations must also address longstanding operational challenges around fragmented data, siloed tools, governance, and operational trust.
According to Manjrekar, the industry is now confronting what he described as the “value gap” between AI experimentation and measurable business outcomes. “A lot of experimentation, very little business value,” said Manjrekar, pointing to infrastructure debt, data debt, and governance debt as key inhibitors to enterprise AI adoption.
The Evolution from AIOps to AgentOps
Traditional AIOps platforms were largely designed to assist operations teams through machine learning-driven correlation, anomaly detection, and incident visualization. While these capabilities improved operational awareness, they often stopped short of meaningful action.
As Manjrekar explained, many organizations still face what he called the “last mile problem.” AIOps platforms can identify incidents and surface dashboards, but operational teams must still convene cross-domain experts to interpret the data, determine the root cause, and decide on remediation steps manually.
AgentOps represents an attempt to close that gap. Rather than simply surfacing information, AgentOps introduces AI agents capable of interacting with operational systems, coordinating workflows, and taking actions based on contextual intelligence. This includes orchestration across IT operations, NetOps, SecOps, cloud environments, and application monitoring domains.
From an enterprise perspective, this shift is significant because modern operational environments continue to grow more distributed and interconnected. AI workloads, hybrid cloud architectures, edge computing, and increasingly autonomous applications are driving new operational complexity that traditional monitoring models were not designed to handle effectively.
Organizations are now trying to understand not only AI itself, but also how to operationalize large numbers of interacting agents across multiple platforms while maintaining governance and control.
Why Unified Agent Control Planes Matter
One of the central themes of the discussion was the concept of a unified agent control plane. According to Manjrekar, many vendors are building agent-based operational capabilities, but most remain constrained by the same fragmented data silos that limited earlier AIOps deployments. Existing tools often operate within isolated operational domains such as applications, infrastructure, networking, or security, making it difficult to create unified operational intelligence across environments.
The challenge becomes increasingly important as enterprises deploy more AI agents that must coordinate decisions across multiple operational domains. Manjrekar described Fabrix.ai’s approach as a domain-agnostic operational intelligence layer capable of federating data and coordinating actions across heterogeneous platforms including APM tools, IT operations systems, infrastructure monitoring platforms, and cloud environments.
A real-world customer example highlighted this challenge. A large enterprise using multiple operational platforms, including Dynatrace, Splunk, CloudWatch, and Oracle Enterprise Manager, struggled to achieve unified visibility because operational teams maintained separate tooling and data ownership boundaries. According to Manjrekar, Fabrix.ai federated operational intelligence across those environments to improve root-cause analysis and remediation workflows.
This reflects a broader trend across enterprise IT. Organizations increasingly recognize that no single operational platform will own the entire operational stack. Instead, interoperability and contextual coordination across multiple platforms are becoming foundational requirements.
Moving Beyond ETL Toward Contextual Federation
Another important topic discussed was the evolution from traditional ETL (Extract, Transform, Load) approaches to what Fabrix.ai calls Entity Context Linking (ECL).
Historically, many operational analytics platforms centralized telemetry into large-scale data lakes or normalized repositories before analytics and automation occurred. However, as operational environments expand and AI agents proliferate, centralizing all telemetry is becoming increasingly impractical.
Manjrekar argued that agent-based systems require contextual structure rather than simply ingesting massive volumes of raw telemetry.
Instead of moving all operational data to a centralized platform, Fabrics.ai uses federation agents that discover data sources, generate metadata schemas, and dynamically create ontology layers. This allows agents to retrieve only the contextual data needed for a given task while leaving the underlying telemetry in place.
From an enterprise standpoint, this model aligns with broader trends around data gravity, sovereignty, and operational scalability. As AI workloads expand, organizations are increasingly looking for ways to minimize unnecessary data movement while still enabling AI-driven operational intelligence.
Governance, Trust, and Operational Control
Despite growing enthusiasm around autonomous operations, enterprise concerns around governance, hallucinations, and operational risk remain significant. Many organizations are still cautious about allowing AI-driven automation to operate without guardrails, particularly in mission-critical environments.
Manjrekar emphasized that operational trust will become a defining requirement for enterprise AgentOps adoption. He outlined several mechanisms designed to provide governance and control, including role-based AI personas, departmental sandboxes, budget and token controls, continuous evaluation modules, and feedback-driven learning systems.
These capabilities reflect a broader market reality: enterprises are unlikely to embrace fully autonomous operations without strong visibility, explainability, and policy enforcement mechanisms.
The conversation also touched on emerging concepts such as the “Mythos” and “Daybreak” frontier models for zero-day vulnerability detection. According to Manjrekar, these technologies highlight the growing convergence between SecOps, IT operations, and threat intelligence workflows.
This convergence further reinforces the need for unified operational intelligence platforms that can correlate operational context, dependency mapping, business impact, and remediation orchestration across domains.
Ecosystem Alignment Becomes Critical
As operational AI matures, ecosystem alignment is also becoming increasingly important. Manjrekar highlighted Fabrix.ai’s expanding partnerships with ecosystem players, including Cisco Systems, IBM, NVIDIA, and Amazon Web Services.
These partnerships underscore another important industry trend: operational AI platforms will increasingly depend on interoperability across infrastructure, cloud, networking, observability, and security ecosystems rather than operating as standalone products.
Ultimately, the discussion highlighted that the future of enterprise operations is moving beyond visibility toward coordinated operational intelligence. The long-term success of AgentOps will likely depend less on AI models themselves and more on how effectively enterprises can unify operational context, governance, interoperability, and trust across increasingly autonomous environments.
For addtional information on Fabrix.ai, please visit their website
