Frequently asked questions

Helvetic AI is an independent Swiss AI evaluation lab that tests AI models automatically: for performance, EU AI Act compliance, FINMA validation, and Swiss language requirements. Every evaluation delivers a HAAS Score across 6 dimensions.

No. You choose from 5 handoff modes: benchmark intelligence (standard, no data needed), API key, Docker on your infrastructure, dedicated hardware on-site, or anonymize-first. In no mode does your data leave Switzerland.

The most affordable entry point is an AI Risk Classification from CHF 3,000. For a full AI Model Evaluation with benchmark results, prices start at CHF 8,000. See all products on our Services page.

An AI Model Evaluation takes 5–10 business days depending on scope. Risk Classification takes about 1 week. FINMA Validation takes 2–4 weeks.

Minimal. In standard mode (benchmark intelligence), you need nothing. We already have the benchmark data. For custom evaluations, you provide an API key. The entire process is designed to minimize your effort.

The Helvetic AI Assurance Score (HAAS) is our composite scoring framework across 6 dimensions: performance, robustness, safety, compliance, Swiss language, and documentation. Each dimension is scored 0–100 with confidence intervals. Details on our methodology page.

Inspect AI is the evaluation framework of the UK AI Safety Institute (MIT License), used by leading AI labs. Compl-AI is the EU AI Act compliance benchmark suite from ETH Zurich, INSAIT. Our system combines both with Swiss-Bench, our proprietary Swiss benchmarks.

Swiss-Bench is our proprietary benchmark suite with 436 scenarios across 11 tasks, testing models in German, French, and Italian on domain-specific tasks. We publish results quarterly as an open-source leaderboard.

(1) A standardized evaluation report with HAAS Scores, gap analysis, and recommendations. (2) Detailed benchmark results, scoring breakdowns, and methodology documentation for independent verification. (3) A findings call for results interpretation.

We are a technical audit lab, not a consulting firm. Our system delivers systematic, reproducible results. No subjective opinions. Entry from CHF 3,000 vs. CHF 200,000+ at Big Four. Every test is repeatable.

Yes. No commercial relationships with any AI model provider. No referral fees, no vendor partnerships, no pay-for-score. Every model is evaluated with the same system and methodology.

FINMA Guidance 08/2024 defines supervisory areas for AI: governance, risk identification, data quality, testing & validation, documentation, explainability, and independent review. Our FINMA Validation evaluates against all areas with 30 FINMA-specific scenarios.

AI hallucinations occur when a model generates plausible-sounding but factually incorrect information: fabricated court rulings, non-existent regulations, wrong financial data. Stanford (2024) found 58% hallucination rate in legal AI analysis. We quantitatively measure hallucination rates as part of the HAAS Score.

Helvetic AI was founded by Fatih Uenal, PhD — with the mission of making independent AI evaluation accessible to Swiss enterprises. Background: PhD (HU Berlin), Postdoc Harvard & Cambridge, MSc Computer Science (CU Boulder), MITx Statistics & Data Science. Based in Bern, Switzerland.

Our methodology is grounded in 40+ peer-reviewed publications from venues including Nature, NeurIPS, ICLR, ICML, ACL, and NAACL. A dedicated scientific article describing Swiss-Bench’s evaluation framework, expert-verified ground truth, and statistical methods is currently in preparation for peer-reviewed publication.