Frequently asked questions

Vigilis assesses your business against the EU AI Act using a deterministic, rule-based engine (no AI in the decision path). Start with a free, anonymous on-screen check; for a full diagnostic and ready-to-adapt compliance documents, request a quotation and we run the assessment for you and email the deliverables. Every finding (applicability, risk classification, obligations) traces to a specific rule and regulation article. No guesswork.

Law firms and consultancies bill by the hour and produce bespoke advice that varies by who you ask. Vigilis runs the same deterministic rule engine every time (identical inputs always produce identical outputs) at a fixed price, with every finding traced to the exact article that requires it. It is not a substitute for legal advice on contested or novel questions, but for standard applicability and risk classification, it removes the variability and cost of hourly billing.

A generic AI chatbot can generate plausible-sounding compliance answers, but it can also invent obligations, miscite articles, or give different answers to the same question asked twice. Vigilis's rule engine is not an AI model, it is deterministic code. AI is only used, where present, to phrase already-computed results in readable language; it never decides your risk classification or which obligations apply to you.

No. The free AI Act check is fully anonymous (no account, no card) and shows your result on screen: an indicative applicability determination, your role(s) under the Act, and an estimated risk level. For the paid AI Act Assessment and Compliance Package you don't create an account either: you request a quotation, we send it, and once it's agreed we run the assessment and email you the deliverables.

The free AI Act check is a short on-screen questionnaire (usually around 10 questions, a couple of minutes) and shows your result immediately. It covers the same ground as the European Commission's official EU AI Act Compliance Checker (whether the Act applies to you, your role, and your risk tier) so you can cross-check the two. For the paid AI Act Assessment there's no long form on the site: you request a quotation, and once it's agreed we send you a structured questionnaire to complete at your own pace. You can mark any question "not sure" rather than guess, it's recorded as a data gap to close, never silently assumed.

The EU AI Act, covered end to end, a comprehensive full assessment plus a free screener. We deliberately build and validate one regulation deeply at a time rather than spreading diligence thin; as the Act evolves and further EU regulation reaches production readiness, it joins the same engine and the same organisation profile.

Very likely yes, in a deployer capacity. Article 3(4) defines a deployer as anyone using an AI system under its own authority in a professional context, that includes companies using third-party AI tools, not just companies building them. Our assessment specifically distinguishes general-purpose tool use from operating a specific AI system, so general ChatGPT/Claude/Copilot use for drafting or research alone won't be over-classified as high-risk deployment. See our article on who the EU AI Act applies to for the full breakdown.

A provider (Art. 3(3)) develops an AI system, or has one developed, and places it on the market under its own name. A deployer (Art. 3(4)) uses an AI system in its operations. Providers carry the heaviest obligations (risk management, technical documentation, conformity assessment); deployers carry lighter but still substantive obligations (human oversight, log retention, worker notification). Our provider vs. deployer article covers this in depth.

Unacceptable-risk practices are prohibited outright under Article 5, the system must not be placed on the market or put into service in the EU at all, regardless of any mitigating measures. There is no compliance pathway for a prohibited practice; it must be discontinued or fundamentally redesigned to no longer meet the prohibited criteria.

The Act entered into force 1 August 2024. Prohibited practices applied from 2 February 2025. General-purpose AI model obligations applied from 2 August 2025. Most of the Act became applicable on 2 August 2026. The high-risk obligations were then deferred by the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force 27 July 2026): systems that are high-risk under Annex III now apply from 2 December 2027, and high-risk AI that are safety components of already-regulated products (Annex I) from 2 August 2028. Timelines can change, we track them, but confirm the current dates for your situation with qualified counsel.

Yes. Tell us how many AI systems you operate when you request a quotation, and we scope the AI Act Assessment (and, in the Compliance Package, the documents) to your whole estate, so each system gets its own role and risk classification.

Vigilis gives you a deterministic, rule-traced assessment of applicability, risk classification, and gaps against the obligations currently in the rule engine. It does not constitute formal legal advice, and does not replace legal counsel for contested interpretations, litigation risk, or novel fact patterns the rule engine has not been built to handle.

€299 for the full assessment, which includes compliance readiness and coverage scores, a full obligation gap analysis, a prioritised action plan, and an audit-ready PDF report. See the EU AI Act product page for full pricing details.

Yes. The free AI Act check is genuinely free and anonymous, no account, no card, nothing stored. It gives you an indicative applicability determination, your role(s) under the Act, and an estimated risk level with key findings, shown on screen. It's intentionally lighter-weight than the paid AI Act Assessment, not a substitute for it.

The AI Act Compliance Package includes everything in the AI Act Assessment plus a full set of ready-to-adapt compliance documents, selected automatically based on your actual role and obligations, not a fixed bundle, so a pure importer gets different documents than a high-risk provider. Documents are built from your answers and tailored with your organisation's real details, and delivered in both PDF and DOCX formats.

Yes. Just request a quotation for the AI Act Compliance Package and tell us you've already had the Assessment, your existing answers and report carry over, so you're only quoted for the additional document work.

See our Terms of Service for the current refund policy. If you believe your report contains an error traceable to the rule engine rather than your own answers, contact us, we investigate and correct rule-engine defects as a priority.

No VAT is charged. Vigilis operates through an Estonian entrepreneur account and is not registered for VAT (we operate below the VAT-registration threshold), so the quoted price is the total amount payable, there is no VAT to add.

The free AI Act check stores nothing, your answers produce an on-screen result and are then discarded. For the paid Assessment, the answers you give in the questionnaire are used only to produce your report and documents, kept only as long as needed to deliver and support that work, and deleted on request.

We share data only with the processors necessary to deliver the service, currently Supabase (EU database hosting) and Resend (email). Payment is by bank transfer, so no card processor handles your data. We do not sell data to third parties. See our Privacy Policy for the full list.

Any data we hold (your request-for-quotation details and paid-assessment answers) is stored in Supabase's EU infrastructure, in the AWS Europe (Ireland) region (eu-west-1). The free AI Act check stores nothing. See our Privacy Policy for the full processor list and safeguards.

Yes. Use our Manage your data page, enter your email and we send you a secure link from which you can download everything we hold about you or delete it. You can also email contact@vigilis.eu. Note that payment records we've already created are kept as long as tax and accounting law requires, then deleted.

No. Your assessment answers feed only the deterministic rule engine that assesses your organisation. They are not used to train any AI or machine learning model.