I work on the layer between people and what they find, and between work and the software that runs it.

A few threads:

• visibility and discovery: how people, businesses, and tools stay findable when AI answers, summarizes, and routes,
• measurement and tracking that survive without cookies, referrals, or dashboards you can trust,
• agent workflows that touch live tools: memory, handoff, orchestration, recovery,
• the data and operational plumbing underneath all of it.

Different surfaces, one layer: the systems that decide what gets found, trusted, and acted on.

Before this, about ten years in technical SEO, analytics, automation, and the infrastructure under digital marketing.

I worked the supply side: crawl logs, index coverage, the measurement layer, the systems that decide what gets seen before a human does. The surface was marketing; underneath were messy distributed systems: APIs that changed without warning, pipelines that quietly drifted, dashboards that lied by omission.

Visibility, measurement, and control were always the real problem. The tools changed — search, then ads and analytics, now AI. The problem didn’t.

Since 2016 I’ve been Head of Data Analytics at Bobdo, a digital marketing agency in Austria, building the automation and agent systems behind client operations.

That gives me a practical view across many real businesses, not one site or one theory: where discovery, tracking, ads, reporting, and automation hit actual constraints. I use it to sharpen the questions, not to expose client data.

What I publish is method, tooling, and aggregate patterns, never client data or names.

I publish when there’s evidence: original-data studies, technical notes, and reproducible methods.

The direction is AI-mediated visibility and work: how systems find, cite, route, remember, measure, and recover. I’m not selling a new playbook or predicting the future of search. I’m building enough instrumentation that the interesting claims can be tested.

I won’t sell you ten steps to beat Google. The problem isn’t shaped like that anymore, and most of the people who say it is are selling the saying.

I’m not training frontier models or timing the AGI clock either. The layer I work on is more boring and more durable: how things get found, how work moves through software, and how you’d actually know any of it worked.