Laying the foundation for agentic AI with resilient 5G connectivity

Even a few minutes of downtime is unacceptable for network managers. It frustrates customers, derails operations, and costs business revenue. Now imagine that outage in a world where agentic AI handles customer service, trouble shooting and remediation workflows. The margin for error disappears.

According to Gartner, agentic AI will solve 80% of customer issues without human intervention by 2029. That means customers and internal teams will expect network services to be ‘always on’. If AI agents are unable to access data or communicate due to a network outage, the impact on productivity and trust will be immediate. To meet the demands of the agentic AI era, enterprises need resilient 5G connectivity that keeps agents available around the clock.

Why AI agents need a resilient network

AI agents are artificial intelligence systems that are capable of performing tasks autonomously or semi-autonomously with human supervision (depending on the deployment). For AI agents to successfully carry out their duties, they must be able to quickly access relevant data sources so they can produce proper outputs or complete tasks. There are several types of AI agents. Some of them include simple functional agents (capable of executing low-level tasks and commands), orchestrator agents (capable of coordinating workflows, assigning tasks to functional agents, and synthesising results), and hierarchical agents (capable of acting as the ‘CEO’ for other agents within a system.) With hierarchical agents, enterprises can implement a multi-agent system. In this system, AI agents work together, almost like agentic coworkers, to satisfy both individual and collective goals.

The nature of AI agents calls for fast communication with data sources, humans, and each other. This communication is only possible with a highly available network. Any interruption in connectivity could create latency in agent responses or put it out of commission altogether. This could have cascading effects on how well an enterprise operates and ultimately affects its bottom line.

For example, a simple reflex agent may manifest as a chatbot, the first point of contact for a potential customer. That agent may be programmed to give simple responses to customers based on their requests. If the customer is asking the chatbot about hours of operation or cost of services — but the chatbot is unresponsive due to a network outage — the customer won’t get the answers they want and may take their business elsewhere.

An airline provides a great example of a multi-agent system. That airline may use such a system to help customers book flights. While one agent provides customers with flight prices, another agent may need to check seat availability on that flight. If the two agents can’t communicate, a customer may book a seat that isn’t available. This is another example where customer frustration could fester and create a loss of business.

Implementing a network built for agentic AI

To prevent these potential frustrations, enterprises can deploy a network solution that can move at the pace of AI. This network should be grounded in the speed of 5G, as well as the flexibility and resilience of an intelligent wireless wide area network (WWAN) solution.

With 5G WWAN, enterprises can deploy intelligent routers that can support agent communication across multiple transport types — Wi-Fi, 5G, LTE and satellite. With this solution, the router will have failover capabilities and transfer communication to the transport type that is most effective at a particular time. This helps maintain network uptime and, in turn, support an agentic AI system that is highly available.

AI workloads can also be data-intensive at times, requiring more network bandwidth. The 5G WWAN solution will also be able to mitigate this reality with link-bonding capabilities. Through link-bonding, network administrators can combine multiple links in their 5G WWAN solution to allow a larger transfer of data and mitigate potential latency issues.

It’s also important to note that 5G provides advantages for enterprises looking to leverage AI at the edge. Through 5G WWAN, enterprises can locally deploy and connect their AI workloads without spending the time and capital that comes with wireline or fibre connectivity. This method ensures locally deployed AI workloads have the benefits of AI at the edge while reducing time to deployment and maximising enterprise resources.

A mutually beneficial relationship between agentic AI and 5G WWAN

As enterprises use 5G WWAN to improve their AI agents, they can also use AI agents to improve their 5G WWAN. For example, network administrators may use agentic AI to assess performance. The administrator could enter queries into an agentic AI-powered virtual assistant. The network manager may ask “How is the network performing?” or “Why was there an outage on July 17th?”. The agent(s) could present detailed responses to both questions, empowering the network administrator to not only address future network issues but also have conversations with C-suite stakeholders.

Network administrators can particularly benefit from a multi-agent system built to support a 5G network. With the hierarchical agent acting as a CEO, it can assign various agents to different tasks to help the network function at its best. If the network administrator needs help configuring a network, the hierarchical agent will assign an agent built specifically for configuration. If the network administrator needs help writing a security policy, the hierarchical agent can assign that task specifically to an agent designed for policy. This system will become especially important for enterprises that may have lean network administration teams and need their staff to spend more time on strategic business initiatives.

The parallel futures of resilient connectivity and agentic AI

As AI agents become more prevalent, outages or downtime will become more costly. Customers will continue to expect an always-available service and could develop less patience for pauses in that service. As a result, the future of resilient connectivity solutions, such as 5G WWAN, and the future of agentic AI are tied together. They create not only a mutually beneficial relationship, but one that may soon be mutually dependent. As an agentic AI system continues to improve network performance, a resilient network will always improve AI agent performance throughout an enterprise.

Image credit: iStock.com/LIORIKI