Michael Korenevsky
Building enterprise AI and predictive tools for high-stakes industries. Ten years in industrial software, the last three shipping AI products to enterprise customers.
Selected work
AI Platform · Oqton · 2025–Present
Took an AI monitoring tool from pilot to five enterprise contracts in five months
Took AMVero from first enterprise pilot to five paying clients in five months — writing the GTM narrative, designing the smart alerting system that eliminated operator alert fatigue, and authoring the deployment playbook that got regulated manufacturers live without disrupting production.
Baker Hughes · Thales · Elos Medtech · 3D Systems · Beehive
Predictive Simulation · Oqton · 2022–2025
Shipped three simulation modules, culminating in the thermo-mechanical solver that made first-time-right manufacturing achievable
Built out the Simulation Suite over three years — Thermal, Mechanical, then the coupled Thermo-mechanical pass that eliminated inter-stage wait times and made first-time-right accuracy viable in serial production. Validated on standard workstations, not servers.
An AI-native PM practice
AI lets me cover more ground in less time and build working prototypes without waiting on development. I can bring something testable to every client and stakeholder conversation instead of a description of what I am imagining.
Discovery & specs
AI synthesized support tickets, Jira history, and customer feedback before writing the AMVero alerting spec — surfaced a threshold conflict that would have shipped a broken alert.
Prototyping
Working mockups in front of operators before any production code was written. They caught a UX problem that would have taken multiple developer sprints to fix post-release.
Market intelligence
Scheduled agent tracks AMVero's regulatory environment, competitor pricing, and new entrants weekly. Flagged competitor pricing shift — data used to propose the move to token-based billing.
Support intelligence
AI read 500 support tickets and found operators were drowning in false alerts, not accuracy problems. That finding redirected the entire Smart Alerting roadmap.
10+ years in enterprise software
A decade of QA before PM means I approach AI product decisions the way a test engineer approaches release certification: assume it will fail, design for the edge case, then validate before shipping.