Complipath helps AI-heavy software companies register, classify and document their AI systems under the EU AI Act.
No account for the check, and the classification runs in your browser.
Complipath is EU AI Act compliance software for companies that build or deploy AI, in Europe and outside it — an AI system register, deterministic risk classification, the obligations that follow, and the evidence behind every decision.
Your first system is free. What Complipath does — and what it does not do — is written out on this page.
What Complipath is, in plain language.
The check is free, and you don't need an account. It runs the same classifier the product runs, citing the article it rests on and the frozen copy of the Act it was derived from.
Give the check your website address and it reads your public pages once, without storing or logging your domain, then asks about your size and what you already publish.
It lists the obligations that apply to that system, each with its application date, then writes out what it does not cover. Built in your browser and sent nowhere, it opens offline.
Keep the result and it becomes your AI system register. The 30-day trial starts with no card.
Every classification names the version of the law it was derived from, and cites the article it rests on.
“Classification derived from Regulation (EU) 2024/1689, Regulation (EU) 2026/1744.”
What the check cannot decide is published, not hidden.
The example run is shown when there is a run to show.
The fine is of worldwide annual turnover, whichever is higher, enforceable since 2 August 2025 and the ban itself since 2 February 2025, with lower tiers of €15m/3% and €7.5m/1%.
Article 99(6)'s SME cap sits on every requirement row in the app.
The high-risk requirements moved to 2 December 2027 for Annex III and 2 August 2028 for Annex I; every assessment here names which text of the Act it was made against, and every answer cites the article it rests on.
Complipath keeps a register of the AI systems your company builds or deploys, classifies each one under Regulation (EU) 2024/1689, and drafts the documentation the Regulation asks for.
Every classification names the article it rests on, who confirmed it and the version of the law it was read against.
It never decides your risk level for you — what Complipath is, in plain language.
Tell us about your AI system
A short set of questions, more if your answers open follow-ups about what it does, who uses it and how it is deployed. Type your domain and we suggest what to register — or name one system yourself.
Get your risk classification
The engine decides the level. Nobody at Complipath does, and no model does either.
Export your documentation
A report you can hand to your team, your customers or an auditor. It says what you answered, on what date and against which text of the law.
Keep it current
When a provision your record cites changes, you are told which systems it touches.
A system can carry that duty and be high-risk at the same time — how the four outcomes work.
Counted across the registers Complipath keeps for itself — internal workspaces only. No customer's register is in these numbers, and none can be: the read is scoped to the internal plan, which is not a plan a customer can be on.
“does not pose a significant risk of harm to the health, safety or fundamental rights of natural persons, including by not materially influencing the outcome of decision making”
“By derogation from paragraph 2, an AI system referred to in Annex III shall not be considered to be high-risk where it does not pose a significant risk of harm to the health, safety or fundamental rights of natural persons, including by not materially influencing the outcome of decision making. The first subparagraph shall apply where any of the following conditions is fulfilled: (a) the AI system is intended to perform a narrow procedural task; (b) the AI system is intended to improve the result of a previously completed human activity; (c) the AI system is intended to detect decision-making patterns or deviations from prior decision-making patterns and is not meant to replace or influence the previously completed human assessment, without proper human review; or (d) the AI system is intended to perform a preparatory task to an assessment relevant for the purposes of the use cases listed in Annex III. Notwithstanding the first subparagraph, an AI system referred to in Annex III shall always be considered to be high-risk where the AI system performs profiling of natural persons.”
Article 6(3), Regulation (EU) 2024/1689 — read out of sources/regulation-2024-1689.html at this build. Not amended by Regulation (EU) 2026/1744 — point (8) inserted paragraphs in Article 6 (sources/provision-status.json). The 2024 wording is in force.
You build or deploy AI in something customers pay for.
Someone will ask you to defend the answer — an auditor, a buyer in due diligence, a customer with a questionnaire.
You need to be able to show who answered, and against which text of the law.
You are established in Europe — or outside it, and covered because the system is placed on the Union market or its output is used there.
It never decides your risk level for you. It shows what the law says, where it says it and what follows.
It does not tick a box for you: every answer carries the article it rests on and the frozen text of the Act it was read against.
It does not tell you what the law will say next year. It dates what it says today, and says when that was read.
Book a 30-minute walkthrough — the reader who is still asking "is this for me" wants a person, not a classifier.
The EU AI Act is in force and its obligations phase in on fixed dates — dates that moved in July 2026. Registration and conformity assessment kept their original date; classification, documentation and human oversight moved to 2 December 2027.
The screening tool nobody registered is the one a reviewer asks about first.
Reconstructing a year of decisions under a deadline costs more than recording them as they happen — before penalties enter the conversation.
The list the Act expects you to have. Every system carries its classification, its obligations and its evidence in one place.
A short set of questions, more if your answers open follow-ups — and the article behind every answer, so the classification is checkable instead of asserted.
What each classification requires, and when. Every date attaches to a classification, because a guessed classification produces confident, wrong deadlines.
The quote, the article and the law version behind every verdict. A person confirms a classification before it counts as evidence.
Complipath drafts your Annex IV technical documentation from your own register, and keeps the evidence behind every classification: the article it rests on, who confirmed it and the version of the law it was read against.
Nothing here is gated to a higher tier. Some of it is not built yet — the list says which.
Your first system is free. What each plan includes, what is built and what is not, and the questions people ask at the price — the pricing page has all of it.
The same answers give the same outcome, every time.
The classifier is a rule engine, not a language model. Every result carries the engine version that produced it.
It can say a use must stop.
Five of the ten prohibited practices in Article 5(1) come back as prohibited, not as a calmer category one tier down. The other five it does not decide, and says so — both lists are published.
It will not exempt you on its own.
The engine may propose that a system is covered. It never proposes that one is not. An automatic exemption is an automatic exit.
We have not tested other tools and we will not claim anything about them. Ask them the same three questions.
The source files the engine reads, each with its sha256 · Check your AI systems - free
What Complipath does
works out which article each system falls under
lists the obligations that follow
gives the date each obligation starts to apply
asks for every point of Annex IV, fills what your register already answers, and shows which points are still yours to add
quotes the provision next to every part
names what could not be determined
What stays a person's judgement
whether Article 6(3) applies to you
whether a practice is prohibited
whether your safeguards are adequate
The record is assembled and cited before a lawyer sees it, so their time goes to the judgement instead of the assembly.
Five things the classification decides, five it hands to a human.
No dates render here, because every date on this page attaches to a classification, and a guessed classification produces confident, wrong deadlines. The classifier decides it in your browser, and each answer cites the article it rests on.
Shorter answers to more questions are on the FAQ page, and what has changed on Complipath, dated and checkable, is on the changelog.
The first system stays free after that, with nothing to apply for and nothing asked in exchange.
The check runs in your browser before you sign up for anything. Or write to hello@complipath.io. That is a person’s inbox, not a form. Complipath is built by Yobel Tzegai in Gothenburg, Sweden.
We measure page views with Vercel Web Analytics. It uses no third-party cookies. Visitors are identified by a hash derived from the incoming request, which is discarded after 24 hours, and no identifier is stored that could follow a visitor to another site. What is collected: the time of the visit, the URL, the referring page, filtered query parameters, city-level location, operating system, browser and device type.