With 6G on the horizon, operators need greater levels of network automation, and AI agents are here to deliver
We have entered the era of agentic AI, and it is here to deliver on the promise of autonomous network operations. Infovista just dropped a new agentic AI framework that is designed to deliver agent-based network automation for telco operators.
VistaAI seeks to move operators from being mere observers of network data to users of the data. In the past, fragmented visibility of network data has caused issues like unexplained service degradations, unresolved network behaviors, and regulatory lapses — leaving the gap between detection and resolution ever so wide.
To turn the problem around, InfoVista injects network operations with agentic AI to act on network data insights. This can potentially change what has always been reactive to predictive — to potentially proactive. The result: operators define the intent and agents execute it.
The solution offers AI agents ranging from “advisory to fully autonomous”, cross-domain data analysis spanning the RAN, core, and transport networks, and a natural language interface for operators to query information conversationally.
“Operators need intelligence that acts. VistAI closes the gap between seeing a problem and solving it,” said CEO, Rick Hamilton.
The rise of intent-based networking
In the mid-2010, a handful of industry groups, led by HPE, came up with the concept of intent-based networking. The idea evolved from the need to move away from manual CLI-based network operations that frequently led to outages to a more efficient and error-free model. The goal was straightforward: design an abstraction framework that reduces the burden of network operators as they struggle with complex cloud networks.
Intent-based networking, when it came, allowed engineers to define intent — or business outcome — as declarative statements. The network converted those into configurations, eliminating the manual steps of coding and execution.
As the cloud infrastructure evolved, IBN stopped being a standalone concept and started being an ubiquitous function. It got baked into Kubernetes in the form of control loops that kept implementations aligned with network policies. It emerged as point solutions that gave enterprises ability to manipulate the overlay and underlay networks through high-level abstraction.
Now as the industry transitions toward sixth-generation (6G) wireless networks, the need for automated orchestration has never been greater. 6G networks need zero-touch orchestration, where high-level operational intents are automatically converted into executable configuration without human intervention.
But existing rule-based systems that power IBN have limitations when applied to the hybrid 5G/6G, edge, cloud and IoT environments. They have semantic gaps and interpretability issues.
Zero-touch management with agentic AI
Networking with agentic AI introduces a new paradigm. Cognitive agents can perform service orchestration, assurance checks, and security analysis of multi-domain networks without human intervention.
A multi-agent framework like Infovista’s allows in-house and third-party agents to collaborate across heterogeneous network environments, enabling faster service delivery.
The agents can make proactive detection by analyzing network states, and even provide anomaly reasoning for a deeper analysis.
Assurance agents can make real-time recommendations and provide remediation steps for faster resolution. On the security front, they can sift through threat intelligence to determine risks and enable rapid response, thus reducing dwell time and blast radius.
Their genuine learning and reasoning capabilities allow them to take individual tasks from intent to execution in one seamless flow, the same way a domain specialist does. They can validate intent objects, decompose tasks into configs, perform roll-outs or roll-backs as required, and limit the scope of errors and failures, ensuring better operations and uptime.
Perhaps, its most significant advancement is the agents’ linguistic adaptability. AI agents come with natural language understanding which makes it possible to define intents in natural language. The agents can still read and translate them into actionable workflows and execute them seamlessly.
However, on the flip side, AI agents can introduce biases that manifest as systemic failures. To avoid that, experts recommend a hybrid approach where complex intents are processed by multi-agent systems, and straightforward, repetitive ones by lightweight ones.
