How it works: the EU AI Act does not switch on in one day. Its duties arrived in stages — and the next stages are dated. This planner starts after classification: if you do not know your category yet, run the free Risk Classification Self-Check first. Already have a category? Select it below, tick only the flags that are actually true, and the timeline at the bottom builds itself — with a printable plan and a calendar file. The wording behind the transparency dates is in the Transparency Disclosure Builder | US Companies Guide.
Your system
These fields only label your plan. They never leave this page. Every status below — already applies, due soon, upcoming — is measured against the plan date you set here.
Your risk category
Choose the one category from your classification. Progress: no category selected
Your classification found a practice the Act bans outright — for example social scoring, manipulative techniques or a banned biometric use. There is no preparation window for a ban: the duty is to stop or not place the system.
Your system is used in an Annex III area: hiring or worker evaluation, credit or eligibility, education, access to essential services, law enforcement, border or migration, or a listed biometric use.
Your system is a safety component of a regulated product, or part of critical infrastructure, on the Annex I track.
Your system's main duties are the Article 50 transparency duties: people must know they are dealing with AI, and AI-generated content must be disclosed and marked.
Everything else: no category-specific application date. Your plan is the dated written record of that reasoning, plus any flags below — flags carry their own dates even on a minimal system.
You have not classified the system. This planner will not guess a timeline from a guessed category — it will tell you exactly what to do first instead.
Flags that add their own dates
Either you provide a general-purpose model, or your product sits on one. GPAI duties run on their own track from August 2025 — and if a foundation model sits underneath your product, recording which model it is, and which duties stay yours, is part of your plan either way.
Generative systems already on the market had a grace period for machine-readable marking of their outputs. That grace period ends on 2 December 2026.
A chatbot, voice agent or assistant that people talk to directly carries the Article 50 interaction notice — on top of any category, including minimal.
The Act has no separate agent category and no separate agent date. Autonomy adds build work inside your category timeline: logging of what the system did and why, and human approval checkpoints where a person must stay in control. The plan lists that work against your category date, not as a legal date of its own.
Your deadline plan
Select a category above
Your dated timeline will appear here as you select a category and tick the flags that apply. Every line names the date, what applies on it, and — for future dates — a clearly labelled suggested preparation window. That window is planning advice, not a legal deadline.
What this planner does not do
This planner does not classify your system, does not perform a conformity assessment, and does not make any date legally certain for your exact system. It turns the category and flags you select into a timeline based on the public application dates of the Act as stated in November 2026 — if the category is wrong, the timeline is wrong with it. Three traps it will not save you from: treating a suggested preparation window as the legal deadline (the work is due when the duty applies, not when preparation starts); assuming a minimal category erases your flags (transparency and GPAI dates apply on top of the category); and treating a future high-risk date as free time (documentation, logging and oversight built in the final month are built badly). Timeline as of November 2026: the Digital Omnibus (Regulation (EU) 2026/1744) moved high-risk duties to 2 December 2027 (Annex III) and 2 August 2028 (Annex I); on 2 December 2026 the marking grace period ends and two new Article 5 bans take effect.
How to use the plan
- Work backwards from the legal dates, not forwards from today. Each future line carries a suggested start — put that start in the team calendar, not just the deadline itself.
- File the plan with its date. A dated timeline is what partners, insurers and enterprise customers ask to see; it also proves the future dates were planned, not discovered late.
- Download the calendar file. It contains only the dated legal lines — preparation windows and undated duties stay in the printed plan, where their reasoning is visible.
- Re-run the planner when anything changes. A new use case, a new market, a new tool connected to an agent, or a rule change restarts the classification — and therefore this timeline.
- Read the reasoning behind your category. The Risk Classification Guide explains the funnel that produced it, question by question.
Want your dates mapped in writing, for your system?
The EU AI Act Classification Report maps your specific system against the Act: risk category with reasoning, the obligations that follow, your deadlines and a step-by-step plan in plain English. $49 one-time, delivered by email. Optional Act Watch is $19/month and re-checks your classification as the rules move. Not legal advice.
Get My Report — $49 Answer the full questionnaireFrequently asked questions
Where do the dates in this planner come from?
From the public application timeline of the EU AI Act as stated on the SkillQuesters EU AI Act pages in November 2026: prohibited practices from 2 February 2025, general-purpose AI duties from 2 August 2025, Article 50 transparency duties from 2 August 2026, the end of the marking grace period and two new Article 5 bans on 2 December 2026, high-risk duties for Annex III systems on 2 December 2027 and for Annex I systems on 2 August 2028 after the Digital Omnibus (Regulation (EU) 2026/1744) moved them. Dates can move again; this planner is information, not legal advice.
What is a suggested preparation window? Is it a legal deadline?
No. A preparation window is this planner's practical suggestion for when to start work — for example twelve months before high-risk duties apply — so documentation, logging and human oversight are not built in the final month. Only the dated obligation lines are application dates stated for the Act; the preparation windows are planning advice, clearly labelled as such.
My category is minimal risk. Why do I need a plan at all?
Minimal risk carries no category-specific application date, but partners, insurers and enterprise customers still ask for the reasoning in writing, and any new use case, market or capability restarts the classification. The plan for a minimal system is a dated record of the reasoning plus the re-check triggers — and any flags you ticked, such as transparency or a general-purpose model underneath, still carry their own dates.
I do not know my category yet. Can this planner classify me?
No. This planner starts where classification ends. If you do not know your category, run the free Risk Classification Self-Check first, or get a written Classification Report for your specific system. A timeline built on a guessed category is a guessed timeline.
What happens to what I type?
Nothing leaves your device. The planner runs entirely in your browser, sends no data anywhere and stores nothing. Close the tab and the inputs are gone — print, copy or download the calendar to keep the plan yourself.
Does the calendar file cover every duty my system owes?
No. The calendar contains only the dated application lines generated from the category and flags you selected. Duties without a single application date — for example keeping logs, maintaining documentation and re-checking after changes — are in the printed plan, not the calendar, and a written classification of your specific system settles what applies to it.
SkillQuesters is independent and is not affiliated with the European Commission. Primary sources: Regulation (EU) 2024/1689 and the European Commission's AI Act pages. This page contains no affiliate links for the service above — the Classification Report and Act Watch are SkillQuesters' own services.