CQI-38 training is a hands-on programme built on the 1st edition of AIAG‘s CQI-38 “AI Vision Inspection Application and Assessment Guideline”, published in 2026. The guideline provides an industry framework for deploying, accepting and managing AI-based vision inspection systems throughout their lifecycle in automotive manufacturing.
Why is CQI-38 necessary?
The use of AI in vision inspection is spreading fast: AI models step in where rule-based systems fall short — in tasks such as feature recognition, defect classification and status evaluation. These systems, however, behave differently from conventional inspection equipment: their performance depends on training data, model version and operating environment, and it can change over time (model drift).
Traditional MSA approaches and rule-based machine-vision validation don’t cover this dynamic behaviour on their own. CQI-38 fills exactly that gap: it defines how the capability, accuracy and reliability of an AI-based vision inspection system are assessed and how they’re maintained over time. The guideline is designed to supplement IATF 16949.
Scope: which systems?
CQI-38 focuses on AI systems that use image data to perform feature recognition, defect classification and status evaluation. Rule-based conventional machine-vision applications and precise dimensional measurement systems are not the guideline’s main focus. The guideline establishes a common language for OEMs, suppliers at all tiers, and vision inspection equipment manufacturers.
Who should attend CQI-38 training?
- Quality teams at manufacturing sites that use, or plan to deploy, AI-based vision inspection systems
- Process and production engineers running machine-vision and AI-based inspection projects
- Quality engineers responsible for the acceptance and validation of these systems
- Engineering teams at suppliers that develop vision inspection equipment and solutions
- SQA/SQD teams and auditors at organizations whose suppliers use AI-based inspection
There are no prerequisites; basic knowledge of the core quality tools (MSA, SPC) makes participation easier.
Training Content
General framework
The purpose and scope of CQI-38 and its relationship with IATF 16949; how AI-based inspection differs from rule-based systems; core concepts: training data, model, classification performance, model drift.
The lifecycle approach
The guideline’s total lifecycle structure: system planning and manufacturing implementation, system and process acceptance, capability maintenance and continual improvement. Expected outputs and decision points at each stage.
Risk-based validation
Setting validation depth according to product and process risk; assessing system capability, accuracy and reliability; defining acceptance criteria.
Change management
Controlling changes involving equipment, AI models, data and operating environments; revalidation triggers; monitoring data quality and the risk of model drift.
Practical exercises
Assessment practice through example scenarios: structuring the acceptance plan for a vision inspection station, revalidation decisions under change scenarios, managing customer-specific requirements.
Learning Outcomes
- Grasping the structure of CQI-38 and its relationship with IATF 16949
- Assessing the capability and reliability of an AI-based vision inspection system
- Planning system and process acceptance on a risk basis
- Managing model, data and environment changes under control
- Strengthening the organization’s AI quality management capabilities
Duration and Participation Options
Duration: 1 day (in in-house programmes, expandable to 2 days depending on scope and depth of practice).
Participation type: Open enrolment or in-house.
Training method: In person, live online, on-demand recording.
Dates: Set according to demand.
Certificate of Attendance
Participants who complete the training receive a certificate of attendance.
- CQI Trainings — which CQI applies to which process, the overall roadmap
- MSA Training — the reliability of measurement and inspection systems
- SPC Training — monitoring process behaviour with data
- IATF 16949 Training — the system framework the guideline supplements
Beyond Training: Implementation Support
After the training, we can provide implementation support: a gap analysis of your AI-based vision inspection system against the CQI-38 framework, structuring the acceptance plan, and setting up the change management system. We plan the scope together based on your needs.
Frequently Asked Questions About CQI-38 Training
Is CQI-38 a special process assessment? No. It is not a special process assessment like CQI-9 or CQI-11; it is a guideline for applying and assessing AI-based vision inspection systems. It includes an assessment approach, but its focus is the inspection system itself, not a manufacturing process.
We have a rule-based machine-vision system — does CQI-38 cover us? The guideline focuses on AI systems that use image data for feature recognition, defect classification and status evaluation; rule-based conventional systems and precise dimensional measurement fall outside the main scope. For teams planning a transition to AI, however, the training serves as a useful guide.
We already do MSA — why is CQI-38 needed on top? MSA assesses the statistical reliability of a measurement system. AI-based systems additionally require managing dynamic elements such as training data quality, model drift, and version and change control; CQI-38 addresses this whole picture.
What should we do if our customer requests CQI-38? Check the customer-specific requirements (CSRs); obtain the guideline and run a gap analysis for your existing vision inspection systems. The training provides the background needed to plan that process.
Do we need to have purchased the CQI-38 guideline for the training? It’s not mandatory; the training conveys the structure and logic of the guideline. At the implementation stage, however, we recommend the organization obtain the guideline from AIAG.