Spirent Luma brings agentic AI to network testing, slashes triage time

Network testing has grown more complex as environments increasingly combine cellular, network slicing and cloud-native architectures. Lab and validation teams now work across multi-vendor, multi-domain stacks, and when failures occur, resolving them can require several domain experts working in parallel. It’s a challenge that Spirent Communications—which was acquired by Keysight Technologies in October 2025—is looking to solve with Luma, its agentic AI solution for network testing and assurance.

“When something fails, whether this is a manual testing or automated testing, predicting the faults and finding the root cause and providing a remediation—that is still a very painstaking, time-consuming aspect of the whole journey,” Anil Kollipara, vice president of product management at Spirent, told Network World. “That problem is an old problem, but we didn’t have a proper solution to be able to comprehensively address that. So, because of that, we introduced Luma.”

Luma is built on a domain-specific knowledge graph, a deterministic rule engine and a multi-agent architecture. The initial release targets Landslide, Spirent’s core network test platform used primarily by carriers, service providers and hyperscalers for pre-production performance testing. The plan is to extend Luma across the Spirent portfolio over time.

How Luma works: Three pillars, multiple agents

Luma is built around three functional pillars: knowledge, test case generation and root cause analysis.

The knowledge pillar handles queries against technical specifications, call flows, product documentation and compliance requirements. The test case generation pillar lets engineers describe a test scenario conversationally and have Luma configure it inside Spirent’s Landslide platform. The root cause analysis pillar processes logs, key performance indicators and packet captures (PCAPs) from a test run to identify the failure point.

Luma executes those workflows through a mixture-of-experts AI architecture. “We’ve built multiple agents within the architecture that are experts,” Kollipara said.

For example, there is an agent that is an expert at processing a PCAP file and understanding what that is. Another agent is an expert at understanding the configuration that is put into the product. Yet another agent is expert at collecting all that information and prioritizing where the root cause is.

“We have about 10 to 12 agents, and we continue to add more, that work together to deliver or execute an entire workflow for the customer, starting from collecting the information like PCAPs, logs and KPIs, and processing all of that information, interpreting the results, and coming up with the root cause,” Kollipara explained.

Determinism is based on domain knowledge

A critical concern with AI is that the LLMs are not deterministic and can potentially hallucinate incorrect information. As it turns out, the way to minimize hallucination is to minimize LLM usage, at least for the Spirent use case.

“We brought in third-party AI platforms, we tried different LLMs, and we quickly, very quickly, realized that the LLM is just part of the solution, and not the entire solution,” Kollipara said. “In Luma, the LLM role is almost just 10% of the whole thing, where it is processing the natural language. Most of the domain knowledge is built into this RAG database, this knowledge graph that we have built in.”

On top of the database layer, Spirent added deterministic rule sets tied to protocol stack behavior.

“We’ve seen cases where you have an issue happening at the subscriber level, there’s a KPI that is wrong, and any ML can find out there is some deviation that is happening,” Kollipara said. “But the root cause is difficult to be articulated just by using machine learning algorithms, because it requires intimate knowledge about the domain, how these are stacked from a protocol stack perspective. Building that rule set in is very important so that you get that deterministic aspect of the output.”

From seven weeks of triage to two minutes with Luma

A core driver of the product is the expertise gap across modern network domains. Kollipara noted that telco experts may not understand cloud native, and cloud native experts may not understand telco. When a failure spans both domains, resolution requires coordinating across teams with different areas of specialization.

Kollipara cited a concrete example from a beta trial. A customer submitted a support ticket that moved through three layers of support before reaching R&D. Between log collection, customer availability and inter-team handoffs, the ticket took seven weeks to resolve. When Spirent fed the same set of files to Luma, the issue was resolved in just two minutes.

Luma for Landslide is the first phase of a broader rollout. Spirent plans to extend the platform to Velocity, its test and lab automation product, and VisionWorks, its live network service assurance platform.

“The platform will be leveraged across multiple product lines and multiple use cases, but the key thing is to have it trained on the domain,” Kollipara said. “It will not be a plug and play where it will be used. We can use Luma directly for a different data set or use cases, but the platform will be leveraged and trained and perfected for that particular workflow.”