Two sides. A.I. users: people and companies that use A.I. models under the law of the place they are in, through the hub, with an account, a card on file and a prepaid wallet, paid per message (an invitation-only pilot for now); a company that uses compliant A.I. pays the same way as a person. And A.I. providers: a provider lists its model on the hub for 5% of what people pay for it and serves the compute, or routes its own platform through the Iron Suit on a monthly plan, which also carries the assessment, the governance documents and the security feature. Each place opens only where its law allows it and its market-access duties are met with evidence.
How much does it cost?
A.I. users pay per message, at the listed rate of the model they choose, from a prepaid wallet: no subscription, and nothing for a request the law where they are forbids, because it is not sent. A.I. providers pay 5% of what people pay for their model on the hub, and a monthly plan (Professional $49/mo, Business $149/mo, Enterprise $299/mo) only when they route their own platform through the Iron Suit. The Pricing page shows both sides.
What AI regulations apply to my company?
The EU AI Act is the world’s first comprehensive AI regulation, applying to anyone placing AI on the EU market or using it in the EU, regardless of headquarters: prohibitions since 2 February 2025, transparency duties since 2 August 2026, high-risk rules from 2 December 2027 (Annex III systems). A breach of its prohibited practices carries fines of up to EUR 35 million or 7 % of worldwide annual turnover, whichever is higher. In the U.S., the voluntary NIST AI RMF provides federal guidance, New York City’s Local Law 144 regulates automated hiring tools, and Colorado’s automated decision-making technology law applies from 1 January 2027. Our platform covers these frameworks so you can assess your obligations in one place. Beyond the assessment, the Iron Suit applies the law of each person’s place at the moment of use, from a Law Catalog researched country by country from the official texts; the Coverage page shows which places are open.
How is this different from enterprise compliance tools?
Enterprise compliance suites are built and priced for large organizations with dedicated compliance teams. Govern All A.I. gives A.I. providers and the teams that build A.I. products the core of that work (risk classification, gap analysis and a prioritized action plan) on a monthly plan, designed for teams without a compliance department. It also enforces the rules at the moment of use: the Iron Suit screens each request and checks each answer against the law of the place the person is in.
Can I try it before paying?
Yes. The Intro Assessment gives you an instant A.I. risk level and your top three compliance gaps with no account. The full inventory, the detailed scoring and the PDF report come with a provider plan you can start at any time. The hub has no subscription: you add prepaid credit and pay only for the answers you receive.
What happens after I complete an assessment?
You receive a breakdown: the EU AI Act risk classification of each A.I. system, your NIST AI RMF score, a gap analysis and a prioritized action plan. The NIST score credits a governance policy only when one is on record: an AI Use Policy generated here, or your own policy recorded in the full assessment. Every provider plan includes a branded PDF report you can share with your board or legal team.
How long does the assessment take?
The free quiz takes about 30 seconds. A full assessment (adding your AI systems and reviewing results) takes 15–30 minutes depending on your portfolio size. The more systems you add, the more valuable the analysis.
Does this only work for LLMs and chatbots?
No. Our regulation engine is model-agnostic. It covers LLMs, image generators, recommendation engines, classification models, speech-to-text, code generation, predictive analytics, detection models, and every other AI system type. Compliance is determined by your use case, industry, and geography, not by the model architecture. If it is an AI system, we govern it.
Why does the hub refuse VPN connections?
The hub applies the law of the place you are really in. A VPN, a proxy, a data centre or Tor hides that place or makes you appear somewhere else, so the hub does not serve those connections and asks you to turn the service off. A home or mobile connection is served, and so are Apple’s iCloud Private Relay and Cloudflare WARP, because both keep your country. Your connection’s address is checked in memory against published lists and never stored. Providers that put the Iron Suit in front of their own products apply the same rule by default.
What is the Iron Suit?
The Iron Suit is Govern All A.I.’s compliance layer between an A.I. model and the people using it. Each request is screened against the law of the place the person is in before any model sees it: a request that law forbids is not sent, and the person is told which law applies and what to change. Each answer is checked against your Personalized Deterministic Guardrails before it reaches your users: an answer that breaks a rule is repaired or withheld, and an answer that cannot be checked in time is not delivered. Deployments list the places your product is offered and can route each place to its own provider. Each check runs on Govern All A.I.’s self-hosted model and adds time to the answer, so today it suits work that can wait, such as drafted emails and reports, better than live chat. Two ways to connect: the proxy (an OpenAI-compatible address) and the validation API.