10+ years working at the intersection of paid growth, marketing analytics, and decision systems for apps and games.
Across subscription apps, gaming, and performance marketing — consistently at the boundary between data and decisions.
AI agents can now access ad accounts, move budgets, and execute campaign changes. That's not a future scenario — it's already happening. The risk isn't that AI is bad at optimization. The risk is that it optimizes faster than humans can catch problems.
The gap I keep seeing isn't in the tooling. It's that the decision logic behind UA — when to scale, hold, cut, investigate, or wait — is usually informal. It lives in a UA manager's head, not in a validated ruleset. That makes automation fragile: recommendation systems inherit undocumented assumptions, and agents act on immature signals.
My focus is on the layer between data and action: signal rights, trust checks, decision trees, guardrails, and human approval workflows. The infrastructure that makes automation safe and useful — not just fast.
Most of my work sits in the uncomfortable middle: the data is available, the team has experience, but the actual budget logic is still partly implicit. My job is to make that logic explicit enough to validate, automate, and improve.
If you're building UA automation, recommendation systems, or connecting AI to your ad stack — and you want to make sure the decision logic is solid first — reach out on LinkedIn.