Know when to revisit
Relevant change notices help identify earlier assessments that may need revisiting.
PRACTICAL OBLIGATION CHECKS FOR AI DELIVERY
Build with AI.Validate what you deliver.
DutAIx validates AI-assisted requirements, designs, code and QA artefacts for alignment with applicable laws and organisational policies—identifying gaps, showing supporting evidence and highlighting what needs human review before production.
For teams using AI to create or develop requirements, designs, code and QA artefacts.
Evidence trace
Recorded example
Project context
AI CV screening & ranking
Recruitment requirement
“Automatically reject applicants below score 60 with no human review.”
Assessment record
Requirements ConflictDesign GapBuild GapQA GapVisible gap & legal basis
Human-review safeguards are not demonstrated.
Human review & corrective action
Review the finding. Correct the requirement. Reassess.
The gap in AI delivery
AI accelerates delivery. Are applicable obligations being addressed?
Requirements, designs, code and tests can progress while relevant obligations remain unaddressed. Examine the work early, before unresolved gaps become part of a production decision.
01 · Requirements
Is the control described?
A requirement needs to describe it.
02 · Design
Is it explained?
A design needs to show how it will work.
03 · Build
What supports it?
Implementation needs supporting evidence.
04 · QA
Has it been tested?
Testing needs to examine the behaviour.
See how DutAIx addresses the gap
Applicability
AI use alone does not establish which obligations apply. DutAIx considers project purpose, regions, organisational roles, intended users and people affected, business domain, risks and relevant facts within its configured coverage.
SEE THE ASSESSMENT AT WORK
Find the gap in the work itself.
One example of AI-assisted delivery: recruitment project artefacts. Switch phases to see selected recorded findings about a missing human-review safeguard.
Requirements module examines stories and requirements, for duties, gaps, conflicts and suggested acceptance criteria or tests.
Recorded fictional project · simplified findings
AObligation
Human oversight in automated recruitment.
Relevant review and intervention safeguards, with regional scope kept distinct.
BEvidence examined
Recorded requirement
“Automatically reject applicants below score 60 with no human review.”
CGap
The requirement explicitly excludes human review. Recorded status: conflict.
Basis: recorded requirement · EU GDPR Art. 22; UK GDPR Arts. 22A–22C. EU AI Act Art. 14 adds future oversight readiness.
DHuman review & action
Review the finding. Add appropriate review safeguards and acceptance criteria, then reassess.
Proposed corrective action · human decision required.
A requirement is intent, not proof of implementation.
Design module examines architecture and design material, for planned controls and relevant passages.
Recorded fictional project · simplified findings
AObligation
Human oversight in automated recruitment.
Relevant review and intervention safeguards, with regional scope kept distinct.
BEvidence examined
design.md · located passage
“Automatically reject applicants below score 60 with no human review.”
CGap
The design excludes human review and does not explain intervention or challenge handling.
Basis: design.md · EU GDPR Art. 22; UK GDPR Arts. 22A–22C. EU AI Act Art. 14 adds future oversight readiness.
DHuman review & action
Specify who reviews a decision, when they intervene and how the applicant can challenge it.
Proposed corrective action · human decision required.
A documented design does not establish that the control is implemented.
Build module examines supported code and configuration, and configured scanner findings, with locations where available.
Recorded fictional project · simplified findings
AObligation
Human oversight in automated recruitment.
Relevant review and intervention safeguards, with regional scope kept distinct.
BEvidence examined
app.py · located code
return {"reject": candidate["score"] < 60}
CEvidence missing
The examined endpoint rejects automatically; review and challenge controls are not evidenced.
Basis: app.py · EU GDPR Art. 22; UK GDPR Arts. 22A–22C. EU AI Act Art. 14 adds future oversight readiness.
DHuman review & action
Review the finding. Provide missing implementation evidence or add the controls, then reassess.
Proposed corrective action · human decision required.
The finding concerns examined code; supported patterns and scanner coverage have limits.
QA module examines test cases and scripts; where configured, screenshots, sampled browser journeys or API conversations.
Recorded fictional project · simplified findings
AObligation
Human oversight in automated recruitment.
Relevant review and intervention safeguards, with regional scope kept distinct.
BEvidence examined
QA-HANDOVER · recorded API reply
“Your application is rejected automatically. No human review or appeal is available.”
CGap
The tested conversation refused human review. Recorded outcome: gap.
Basis: QA-HANDOVER transcript · EU GDPR Art. 22; UK GDPR Arts. 22A–22C. EU AI Act Art. 14 adds future oversight readiness.
DHuman review & action
Correct the workflow and test human-review and challenge routes. Reassess the resulting evidence.
Proposed corrective action · human decision required.
This is one recorded probe; sampled observations do not prove every behaviour.
Recruitment is one example of the assessment approach. Use modules independently or together; each phase keeps its own evidence, obligations and result.
Explore the phase checks in detail →COMPLIANCE EXPERTISE, APPLIED
Your compliance expertise.
Built into AI delivery.
Your compliance team or authorised personnel can review and publish obligations drawn from applicable laws and organisational policies—giving delivery teams a consistent, reviewed basis for assessment.
Project context determines which configured obligations are relevant. The material from each selected phase shows what has been addressed and where gaps remain.
A REVIEWED BASIS FOR DELIVERY
Your compliance team or authorised personnelReview obligations. Publish updates.
REVIEWED & PUBLISHEDApplicable laws + organisational policies
PROJECT CONTEXT
Regions · roles · business context · risks
RequirementsDesignBuildQA
Relevant obligations. Selected phases. A consistent assessment basis.
CONFIGURABLE MONITORING
Configure automatic source monitoring, review changes and extend regional coverage as your organisation’s needs evolve.
- 01Monitor changes
- 02Review obligations
- 03Publish updates
- 04Assess relevant projects
Ask the Lab
A conversational route to supported project information and actions, with confirmation steps for supported admin actions.
Take clear findings into your next delivery decision.
See the applicable obligations, inspect supporting evidence and understand what needs attention. Explore a practical walkthrough using fictional project material.
- Relevant obligationsA defined assessment scope
- Evidence-backed findingsReferences, gaps and uncertainty
- Proposed next actionsA basis for human review
Request a walkthrough
DutAIx provides assessment support, not legal advice or compliance certification. Results depend on configured coverage, project facts, supplied evidence and human review.