Shipping Luma for Landslide

Spirent's Active Test Automation (ATA) portfolio had accumulated years of observability data from tier-one telecom operators, but the insight extraction remained manual. Engineers spent hours correlating logs, metrics, and traces to diagnose issues that showed recurring patterns. The commercial opportunity was clear: operators would pay for faster, more reliable root cause analysis. The organizational challenge was harder. Spirent had never shipped an AI-native product. The R&D organization was skeptical of AI hype, and product leadership wanted proof of commercial viability before committing resources.

- Audited the full ATA portfolio to identify the five most common use-case patterns where AI could add measurable value - Narrowed the initial product scope to three patterns: automated root cause analysis, anomaly detection, and resolution recommendation - Chose to ship into an existing sales channel (Landslide) rather than pursue a new-segment motion where commercial viability was unproven - Built the agentic architecture with explicit reasoning traces so operators could understand why the system reached its conclusions - Established feedback loops so the system improved with each resolution

Shipping AI into production requires narrowing scope ruthlessly. We started with five use-case patterns and shipped three. The discipline to cut was harder than the technical work. Operators don't trust black boxes—showing the reasoning chain was as important as the answer itself. Selling into an existing channel meant we could prove value without betting the company on a new GTM motion.