01
Product work
Turned an alert-heavy AI pilot into a trusted product used by five enterprise customers
The challenge
Turn an AI monitoring pilot that operators did not trust into a product customers could buy, install, and rely on.
My role
Led product strategy from customer discovery through requirements, pricing, launch, and deployment. Led an international group across 6 engineers, 2 designers, 2 marketing partners, 3 sales partners, and 5 application engineers.
Selected customersBaker Hughes · Thales · Elos Medtech · 3D Systems · Beehive
03 / Product decisions
The decisions that made the product useful
Each choice starts with the customer problem and ends with the practical result.
Decision 01
Signal over noise
Outcome: ~90% fewer layers for engineers to review
Operators were reviewing alerts on roughly 3,000 of 6,000 layers, so they could not tell which events needed action. I set rules that required the same issue across multiple layers before raising an alert. That cut the review set by about 90% and let engineers focus on genuinely critical events.
Decision 02
Roadmap judgment
Outcome: better operator visibility and protected uptime
In the closed on-premise environments we served, a few known users could work without SSO. The more urgent customer need was knowing when the system was unhealthy, so I deferred SSO and shipped Diagnostics first. That gave operators a way to spot and fix reliability problems before production was disrupted.
Decision 03
Enterprise adoption
Outcome: self-deployment in one day, not days of support
Command-line installation made customers depend on days of technical support before they could see value. I prioritized a guided installer and documentation so they could self-deploy in one day and begin using the product without waiting on our team.
“We've seen a 98% reduction in engineering review time per build, allowing our team to focus on more critical tasks. This, combined with an 18% reduction in scrap costs, has delivered a powerful return on investment.”
