Prerequisites

Before you begin, make sure you have the following:

A valid business email address (personal email addresses are accepted on the Free plan)
Basic information about the AI system you want to assess: what it does, who uses it, and what domain it operates in
Owner, Admin, or Compliance Officer access within your organisation (required to invite team members and publish documentation)

No technical background is required to complete your first risk assessment. If you plan to connect MLOps tools or use the API, refer to the Integrations and API Reference pages.

Step 1: Create Your Account

Navigate to app.aikraft.eu and click Get Started Free.

Enter your work email and choose a password, or sign in with Google Workspace.
Verify your email address via the confirmation link sent to your inbox.
On the onboarding screen, select your role: Compliance Officer, Technical Lead, or CTO / Executive. This personalises your dashboard but does not restrict any features.
Select your organisation’s primary country of operation. Aikraft is hosted in the EU (AWS Frankfurt region) and this selection is used for jurisdiction-specific guidance.

You will land in the Aikraft dashboard with a prompt to add your first AI system.

Step 2: Add Your First AI System

Click Add AI System in the top-right corner of the dashboard. You will be prompted to fill in four fields:

Field	Description	Example
Name	Internal name for the system	`Candidate Screening Model v2`
Description	Plain-language description of what the system does	`Ranks job applicants based on CV content and interview scores`
Purpose	The business objective the system serves	`Reduce time-to-hire for HR team`
Domain	The sector or operational context	`Human Resources / Employment`

Click Save and Classify. Aikraft will use this information to pre-populate the classification questionnaire in the next step.

Step 3: Run Your First Risk Assessment

After saving your system, you are automatically taken to the Classify module. The questionnaire contains between 10 and 15 questions depending on the domain you selected. Questions cover:

Who the system’s decisions affect (individuals, groups, public services)
Whether the system operates autonomously or with human oversight
The type of data the system processes (biometric, financial, health-related, etc.)
Whether the system’s output is binding or advisory

Answer each question based on how the system is actually deployed, not how it was designed. Aikraft flags contradictions if, for example, you indicate human oversight but also indicate that decisions are implemented automatically.

At the end of the questionnaire, click Run Classification. Results are returned in seconds.

Step 4: Understand the Dashboard

Once classification is complete, your system appears in the main dashboard with a risk tier badge (Unacceptable, High, Limited, or Minimal). The dashboard has three main sections:

Classify

Shows the risk tier, the Annex III category (if applicable), and a summary of the factors that drove the classification. You can re-run classification at any time if the system changes. See AI System Risk Classification for a full explanation of the logic.

Document

Shows the status of your Annex IV technical documentation. For high-risk systems, documentation is mandatory under EU AI Act Article 11. The AI copilot generates a first draft automatically. See Technical Documentation Generator for a step-by-step walkthrough.

Monitor

Shows the real-time compliance health of your system: data drift alerts, incident reports, and regulatory change notifications. Monitoring requires at least one classified system and is configured separately. See Setting Up Compliance Monitoring.

What to Expect After Your First Run

Free plan: You can fully classify, document, and monitor one AI system. Results are stored indefinitely. Exporting documentation as PDF requires a Starter plan or higher.
Invited team members receive an email with a link to join your organisation’s workspace. You can set role-based permissions under Settings > Team.
Audit trail: Every classification run, document edit, and monitoring alert is logged with a timestamp and user identity. This log is available under Settings > Audit Log and can be exported for regulator submissions.

If you encounter any issues during setup, contact support at aikrafteu@proton.me or use the in-app chat widget.