Platform / Early exposure screen
Know what to check next—and what needs human review.
A short screen should not pretend to settle a complex legal classification or discover every operational risk. EU AI Fit starts proportionately, then gives workspace teams a deeper, retained assessment when the system's context, effect or uncertainty justifies it.
Product demonstration · 1 minute 57 seconds
Guided AI exposure assessment
See unknown answers become evidence questions, selected answer conflicts receive review, and authorised people decide the next step. Includes labelled process illustrations. All organisations, people and systems shown were created for demonstration, and no client data appears.
Watch on its own page with the full transcriptRead the video transcript
What should you do when an assessment answer is unknown? EU AI Fit helps your team find the next useful question.
In this scenario, Northstar Recruitment is reviewing an application ranking tool. The supplier says it has tested the product, but the team has not seen results for its intended use.
Rather than guess, the contributor records testing as unknown. The workspace asks which tests cover this use, what acceptance criteria were applied, and where the results can be found.
It explains why that matters, suggests useful evidence, and identifies a colleague to consult. That is a suggested contact, not an automatic assignment or email.
The reviewer can request evidence. The finding stays open, and My work links the team back to the retained question. Missing information is not proof that a system has failed, or that it is safe.
Selected conflicting answers also need attention. If the assessment describes a material decision about people, but says nobody is affected, the workspace flags that conflict.
Correct the answers in a new assessment before approval. Earlier versions remain in history. These checks cover selected conflicts, not every possible inconsistency.
Contributors prepare the facts. An Owner or Administrator reviews the findings and approves the exposure map. The contributor cannot approve their way past that review.
After the map is approved, the next action leads to any outstanding regulatory role review, then classification. Each needs its own decision. Map approval is not operational readiness.
Defined, versioned rules help you investigate uncertainty and keep a reviewable record. They do not discover every risk or certify compliance. Visit our website to explore the guided workflow.
Why this matters
Most teams begin with a product name. The Act begins with purpose and context.
The same model can support a low-impact writing assistant, a workplace decision or a safety function. A useful first step therefore asks what the system does, who it affects, how its output is used and where it is deployed.
EU AI Fit separates obvious scope and transparency signals from deeper questions about affected people, data, fairness, accuracy, security, human oversight, suppliers and operation. Uncertainty remains visible instead of being converted into a reassuring score.
You do not need to guess to finish an assessment. In the workspace, an explicit unknown answer brings up a question to investigate, why it matters, useful evidence and a suggested colleague to consult. The suggestion does not assign work or contact that person automatically.
Who it is for
For the people supplying facts and reviewing what remains uncertain.
- Teams discovering AI use across the organisation
- Product and procurement owners considering a new system
- System owners and reviewers examining operational exposure
- Advisers scoping an initial client conversation
How the work moves
From an initial use case to a reasoned next step.
- 01
Screen the use
Capture enough purpose, context, affected people and decision influence to identify the principal EU AI Act signals.
- 02
Work through the unknowns
The deeper workspace assessment examines ten exposure areas. Save uncertainty, follow the evidence prompts and correct flagged answer conflicts through a new assessment version. Earlier versions remain in history.
- 03
Submit for human review
Contributors can prepare the record; an Owner or Administrator reviews findings and approves the map. A request for evidence remains open work. Detected answer conflicts must be corrected before approval.
- 04
Follow the outstanding decision
My work links to the next retained finding. Once the map is approved, it directs the team to outstanding role review, then classification. Neither approval nor operational readiness follows automatically.
In this scenario
The supplier says it has tested the recruitment tool. The team has not seen the results.
In this scenario, a people team is assessing a plug-in that ranks applications. It cannot yet confirm which tests cover its intended use.
The contributor records testing as unknown. The workspace asks which tests cover the intended use, what acceptance criteria were used and where the results can be found. It suggests dated test evidence and a technical or assurance lead to consult—not a conclusion that the tool is safe or has failed.
The reviewer requests evidence, so the finding remains open and My work returns the team to that question. When evidence is available, an authorised reviewer records the decision. If the answers also describe material decisions about people but no affected people, that detected conflict requires a corrected assessment; it cannot be waived away.
Approval of the corrected exposure map does not finish the assessment. The next step is any outstanding regulatory-role review, followed by classification. Credible risk hypotheses remain separate work for formal review.
The team knows which facts are missing, where to look and who can approve the record. It has a practical route forward without inventing an answer or treating a questionnaire as a legal conclusion.
What is retained
Keep the answers, open questions and review decisions together.
Email the free-screen answers, or continue in the workspace with retained findings, evidence prompts and separate approval decisions.
- Dated public-screen answers
- Versioned workspace answers, findings and evidence prompts
- Finding resolutions, rationale and approval
- Suggested risk hypotheses selected for formal review
- Review date, reassessment history and outstanding next step
Source basis
The official material behind the workflow.
Source review updated 31 August 2026. Read our editorial and regulatory review policy, check the current source and obtain qualified advice for your circumstances.
- Regulation (EU) 2024/1689
The official consolidated legal text, including the 2026 amendments, and starting point for every assessment.
- Article 6 and Annex III
Official classification rules for high-risk AI systems and the documented-assessment requirement.
- Article 50 transparency guidance
Commission guidance on transparency obligations applying from 2 August 2026.
Next step
Start proportionately, then deepen where the facts lead.
Run the free screen, keep your answers and use the workspace to turn material exposure into reviewable decisions and owned work.
Start the exposure check