Independent AI due diligence for investors. EU AI Act readiness for operators. From a builder-operator who ships production AI systems — analytics engines, tender intelligence, AI-native SaaS — and then vets them for others. No product to upsell into the assessment. No vendor agenda.
Book a 30-min call See how I help
I sit on both sides of a table most people keep apart: I build production AI systems, and I vet them. Through EM Consulting, I design and ship AI analytics engines, tender & procurement-intelligence platforms, and AI-native SaaS products end to end. Then, for investors, I run independent due diligence on AI companies before a round closes — and for operators, I map real exposure under the EU AI Act by reading the primary text, not the headlines.
The value on the assessment side is the independent read: no product to upsell into it. Twenty-plus years across regulated, cross-border environments. Based in Vienna.
Explainable risk-scoring across thousands of structured records — reports, filings, transactions. Every flag comes with a plain-language reason, not a black-box score. Built to run inside your own infrastructure; your data never leaves your environment.
AI systems that turn thousands of tenders and filings into a single, queryable, risk-ranked view — cross-border, cross-currency, cross-language. I've built and operate platforms doing exactly this at production scale: AZTender (Azerbaijan) and Veritender (EU / institutional).
From spec to production: architecture, data pipeline, model integration, deployment. I design and ship AI-native SaaS platforms end to end — the same standard I hold vendors to when I'm the one doing due diligence.
An independent read on one AI company before you wire — or before your investor does. Tech reality vs pitch claims, the "AI moat" stress-test, team & build risk, data and evaluation discipline, EU AI Act exposure, and the red flags that survive a polished deck.
The 7 May omnibus did not delay the AI Act — it split it into four timelines. I re-derive yours against the amended text: prohibitions (live), GPAI (live since Aug 2025), transparency (still 2 August 2026), high-risk Annex III (now December 2027). Then classify your systems, score the gaps, and hand back a prioritised 30-day plan.
Most AI advice is sold by people who also sell the remediation — or who have never shipped the thing they're assessing. I do both jobs that are usually kept apart: I build production AI systems, and I vet them. Twenty-plus years across regulated, cross-border environments taught me where AI risk actually hides — in the eval suite and the org chart, not the slide. The independent read is the whole product. Nothing is being upsold to you underneath it.