3.4 ยท The Approval Workflow โ From Idea to LiveDownload
Build Your Approval Workflow
โฑ 10 minCourse 03
Use this section's key principles to build or upgrade your organisation's AI approval workflow. The framework below summarises the Eight Gates model and the key questions at each stage.
The Eight Gates โ Quick Reference
Gate 1: Problem Definition
- โIs the problem clearly documented?
- โIs the current-state cost quantified?
- โIs the AI hypothesis specific and testable?
Gate 2: Data Feasibility
- โAre data sources identified and quality assessed?
- โIs the legal basis for data processing confirmed?
- โAre there any data gaps that could block the project?
Gate 3: Ethical Review
- โFive-Dimension Framework score above 12?
- โNo legal or organisational red-line violations?
- โRisk mitigation plan for any weak dimensions?
Gate 4: Build vs Buy
- โBuild/buy analysis documented?
- โVendor risk assessed if buying?
- โCustomisation requirements realistic?
Gate 5: Pilot Design
- โPilot scope and duration documented?
- โSuccess metrics defined in advance?
- โBaseline measurement methodology in place?
Gate 6: Risk Assessment
- โAll material risks identified?
- โMitigations documented?
- โResidual risk accepted by appropriate authority?
Gate 7: Stakeholder Sign-Off
- โAll required approvals obtained?
- โWorks Council consulted if legally required?
- โBoard sign-off for high-risk AI?
Gate 8: Post-Deployment Monitoring
- โPerformance metrics defined?
- โMonitoring cadence and escalation triggers documented?
- โNamed owner assigned for ongoing oversight?
