What you bring
Access to your AI system and supporting documents for the agreed checks. Together, we confirm its intended use, your organisation's role and the applicable requirements.
Check your AI system against the EU AI Act.
Connect your AI system to Baselex to analyse its behaviour alongside your documents against applicable EU AI Act requirements. See which checks are met, where the system falls short and what evidence is missing.
AI GovernanceEU AI Act ModuleAvailable for demos and pilots.
What Baselex checks
System behaviour and supporting documents against the requirements in scope. See checks that are met, behaviours that fail a check and gaps that need more evidence.
What you receive
Findings that link each requirement to observed behaviour, documents and the check performed. Your team can inspect the evidence, resolve issues and make the decision.
System behaviour and documents. One review.
In this fictional connected test, the chatbot shows no AI notice at first interaction. The supporting documents also lack evidence of staff AI literacy measures.
Next: resolve the notice finding, repeat the connected test and provide the missing evidence for review.
This fictional case concerns a company deploying a vendor chatbot. Duties depend on the organisation's role and the system's use; these selected checks do not amount to full AI Act clearance.
EU AI Act ModuleFictional example
What still needs attention before we roll out this chatbot?
An EU company deploying a vendor chatbot that answers routine product and order questions directly to customers.
Pending
Review pending. Resolve the AI notice gap and supply evidence of staff AI literacy measures before the team's internal approval decision.
Selected readiness checks Open a finding to follow its evidence trail
Intended use and roleThe supplied brief identifies the use and the company as deployer. Scope confirmed
- Regulatory referenceEU AI Act Article 3(4) (opens the official regulation in a new tab)
The review starts with who uses the system and for what purpose. A deployer uses an AI system under its authority; this establishes scope, not a compliance pass.
- Exact fictional sourceSystem Use Brief.pdf
Section 2 | Intended use and roles | page 2
"Our EU customer-support team uses the vendor chatbot for routine product and order questions. The vendor provides the system; our company deploys it."
- Baselex assessmentScope confirmed
The described use and deployer role are documented. This selected check is not a full risk classification; high-risk requirements are not assumed for this routine support use.
AI notice at first interactionThe AI notice is off, and the opening message does not identify the AI. Not met in the connected test
- Regulatory referenceEU AI Act Articles 50(1) and 50(5) (opens European Commission guidance in a new tab)
For direct interaction, the provider must design the system to inform people they are interacting with AI, unless this is obvious. The notice must be clear and accessible by the first interaction.
- Exact fictional sourceChatbot Setup.pdf
Section 4 | Customer-facing interface | page 6
"AI notice: disabled. Display name: Alex. No AI label or other AI-identifying interface cue."
Connected Chatbot Test.txtControlled test | Messages 1 and 2 | First interaction
"Customer: Where is my order? Alex: Hi, I am Alex from support. Let me help you with that."
- Baselex assessmentNot met in the connected test
The documented setup and connected test show no AI notice or obvious AI cue. The selected deployment control fails the internal approval check. Article 50(1) is a provider obligation; this is not an automatic finding that the deployer has breached that provision.
Staff AI literacy measuresA briefing is planned, but implemented measures are not evidenced. Need evidence
- Regulatory referenceEU AI Act Article 4(1) (opens European Commission guidance in a new tab)
Providers and deployers take measures to support the development of staff AI literacy, considering their experience and the context of use. No specific proficiency level or training certificate is assumed.
- Exact fictional sourceRollout Plan.pdf
Section 5 | Staff readiness | page 8
"The support-team AI briefing is pending. No record of implemented AI literacy measures is included in this review pack."
- Baselex assessmentNeed evidence
A planned briefing alone does not show which measures have been implemented for the staff operating this chatbot. The supplied pack does not yet evidence this check.
What happens next?
Update the notice and supply the missing evidence. Agents recheck the selected findings; the accountable reviewer decides whether to approve under the agreed scope.
Real regulatory references; fictional documents, connected test and findings. These selected checks are not a full risk classification, a complete EU AI Act assessment, certification or legal approval. This website does not perform an assessment or authorise an action.
Start with one AI system.
In a demo, see how Baselex connects to your AI system and checks its behaviour alongside supporting documents. Before a pilot, agree the applicable requirements, system access, evidence sources, review roles and what a useful result would look like.
Baselex can recheck affected questions when configured system facts, regulations or evidence change. Your team still decides what the findings mean for approval.
Findings show the result of the checks in scope. They are not certification or a claim that every applicable obligation has been met. Your team retains responsibility for legal interpretation and required approval. Read about scope, data and accountability.
Connect one AI system. See what needs attention.
Tell us about its use and your team's review. See how Baselex checks system behaviour and documents, then discuss a focused pilot.
Request a demo