Skip to main content

Compliance assistant (AI support chat)

The floating chat bubble in the bottom-right corner of every dashboard page opens the compliance assistant. It answers questions about the EU Cyber Resilience Act and about using CVD Portal, always citing the official documents it drew from, and it guides you through your setup step by step until you are done.

What it knows about your account​

The assistant reads your live progress each time you open it and again with each message, and shows it in the panel header:

  • Portal setup: the five setup steps (company profile, reporting channel, submission alerts, first report, team) with how many are done.
  • CRA product compliance: how many of your products have a complete assessment, and what each product still needs.
  • Article 14 reporting: whether the minimum is met, how much of the September 2026 readiness is done, and the CSIRT that your establishment settings resolve to under Article 14(7). After you change those settings, ask the assistant to check again.
  • Urgent items: an overdue Article 14 deadline or a breached acknowledgment window always takes priority over setup guidance.

Ask "What should I do next?" at any point and it recommends the single most useful next step, with a link to the exact page. The suggested questions under each answer follow your progress as well.

Questions and citations​

For any question about the regulation (obligations, deadlines, Annex I, Article 14, standards) the assistant answers from the official knowledge base, which contains the CRA text, EU guidance and the product documentation. The assistant reads the source passages itself and quotes the article wording for an obligation. Sources are listed under each answer. If the passages do not answer the question, the assistant says so. If the assistant is unavailable, it shows the matching documentation pages and does not give an answer. It is available on every plan.

Each answer goes through a rule check after it is written. The check looks for an article number from the 2022 proposal of the CRA, for example "Article 11" for the reporting of actively exploited vulnerabilities, which is Article 14 in Regulation (EU) 2024/2847. It also looks for a wrong statement of the support period or of the Article 14 reports. When it finds one, a note under the answer names the sentence and the correct article. The check does not change the answer. A sentence that names the 2022 proposal is not flagged. An article written directly after "not" or "wrong article" is not flagged either, for example "Article 14, not Article 11".

Actions it can run for you​

On the Compliance plan and above, account admins can ask the assistant to do certain steps directly from the chat. The assistant never acts on its own: it shows a confirmation card describing exactly what will run, and nothing happens until you click Run this. Available actions:

  • Analyze documents. Attach up to 5 files (PDF, image or text, 5 MB each) with the paperclip, and the assistant can analyze them, detect the product they describe, and map them to CRA controls as evidence.
  • Create a product. After an analysis, ask it to create the product and the analyzed documents are attached as evidence, ready for review in the product workspace.
  • Draft an artifact. Ask it to draft a specific technical-file artifact and, once confirmed, the draft appears in the product's Requirements tab with status Generated for your review and approval. Drafts count against the same AI limits as drafting in the workspace.

If your plan does not include an action, the assistant still explains the step and how to do it manually.

Privacy​

Each chat is kept encrypted for 30 days, so that a second AI model can check the answers for accuracy, and is then deleted. The support team can see the results of that check. Do not include vulnerability details or personal data in chat messages. Attached files are only uploaded when you confirm an analysis action, and then follow the same storage rules as evidence uploaded in the product workspace.