Certification of Industrial AI Training Providers
Requirements for a centre, a university or a training department to evidence that what it teaches about industrial AI reaches the job. The classroom is not what is assessed: the shop floor six months later is.
Complete normative text, published openly and free to read.
The Problem: Training That Is Liked and Changes Nothing
In industrial technical training, student satisfaction and shop floor usefulness correlate poorly. A course can score 9 out of 10 and not change a single factory decision. That is why this standard moves the measuring point: from the end of the course, to six months later on the job.
Excellent on fundamentals: models, metrics, libraries. It is what the market is full of.
Does not solve: dirty industrial data, the control layer, drift, or what to do when the model is wrong on a running line.
Delivered by the manufacturer on its own tool. Very operational and very concrete.
Does not solve: independent judgement. It teaches how to use a tool, not how to decide whether that tool is the right one.
Certifies the provider and each programme, with mandatory industrial content and transfer measurement.
Adds: 16 programme requirements, shop floor experience required of faculty and 6 indicators with thresholds.
IRIS-EDU 21001:2026 Normative Text
The complete body, in the open. A company buying training can use Annex A as a specification, and a provider can self-assess before deciding whether certification pays off.
Contents
- Scope
- Normative references
- Terms and definitions
- Context of the organisation
- Leadership
- Planning
- Support
- Operation
- Performance evaluation
- Improvement
- Annex A — Training programme requirements
- Annex B — Faculty requirements
- Annex C — Transfer indicators
- Annex D — Catalogue 21001–21050
1 · 2 · 3 — Scope, References and Definitions
1 Scope
This standard specifies the requirements for a training provider to establish, implement, maintain and improve a management system for industrial artificial intelligence training, and for its programmes to be recognised as conformant.
1.1 Who it applies to
It applies to universities, business schools, vocational training centres, private academies, in-house training departments of industrial companies, and technology suppliers who train on their own products.
1.2 What is certified
Two things are certified together and neither separately: the provider, through the management system of clauses 4 to 10, and each training programme within scope, through the Annex A requirements.
1.3 The principle governing this standard
Training is not assessed by what happens in the classroom but by what happens six months later on the job. The Annex C indicators are the core of the standard: a provider with excellent syllabuses and zero transfer is not conformant.
1.4 Exclusions
This standard does not accredit official qualifications nor replace the competent education authority. It does not certify the people trained, who are certified through IRIS-PRO.
2 Normative references
The following documents are referred to in whole or in part and are indispensable for the application of this standard.
- ISO 21001 — Management systems for educational organisations.
- ISO 29993 — Learning services outside formal education.
- ISO/IEC 17024 — Requirements for bodies certifying persons.
- European Qualifications Framework (EQF) — Reference levels.
- IRIS 14001 — Industrial AI software conformity standard.
- IRIS-PRO — Personnel certification scheme.
- Regulation (EU) 2024/1689 — Artificial Intelligence Act.
3 Terms and definitions
3.1 training provider
Organisation that designs, delivers or assesses industrial AI training programmes, regardless of its legal form.
3.2 training programme
Structured set of content, activities and assessments with a declared duration, entry profile and learning outcomes.
3.3 transfer
Effective application of what was learned to a real task in the trained person's job. It is the object of measurement of indicator K1.
3.4 learning outcome
Verifiable statement of what a person can do on completing the programme, expressed as observable action.
3.5 real setting
Operating industrial installation, or a replica reproducing its data, rhythms and constraints with fidelity sufficient for the decisions learned to hold on the shop floor.
3.6 technical instructor
Person delivering the industrial technical content modules. Subject to the shop floor hours requirement of indicator K4.
4 to 10 — Management System Requirements
4 Context of the organisation
4.1 Understanding the context
The provider shall determine the industrial sector it serves, the real profile of its students and the distance between the training it offers and the competences demanded by the industry around it.
4.2 Interested parties
At least the following shall be identified: students, the companies employing or sending them, the faculty, education authorities and, where the programme is a reskilling one, the legal representation of workers.
4.3 Scope
The scope shall name each programme, its delivery mode, its duration and the required entry profile.
4.4 Management system
The provider shall establish and maintain the processes needed to design, deliver, assess and improve its programmes, and their interactions.
5 Leadership
5.1 Leadership and commitment
Management shall ensure that teaching resources, practice data and access to real environments are available before committing to an edition of the programme.
5.2 Policy
A training policy shall be established including the commitment to shop floor applicability and to keeping technical content current.
5.3 Roles
A programme owner shall be designated for each programme in scope, with authority to suspend an edition that does not meet the conditions of clause 8.
The programme owner shall not be the same person who signs off the final student assessment on that same programme.
6 Planning
6.1 Risks and opportunities
The risks that prevent transfer shall be identified: outdated content, absence of real data, faculty without shop floor experience, poorly filtered entry profile.
6.2 Objectives
Measurable objectives aligned with the Annex C indicators shall be set, with their baseline and target.
6.3 Planning of changes
Every substantial change to the technical syllabus shall be planned and recorded, with its reason and date.
7 Support
7.1 Resources
Practice data, the laboratory environment and access to installations or replicas sufficient to meet requirement P.4.1 shall be secured.
7.2 Faculty competence
Faculty competence shall be evidenced. For technical modules, evidenced shop floor experience is required in accordance with indicator K4.
7.3 Awareness
Faculty shall know the IRIS principles and the rationale behind the Safety AI and human impact requirements they deliver.
7.4 Communication
Before enrolling, students shall receive: learning outcomes, assessment criteria, entry profile and what the programme does not cover.
7.5 Documented information
Syllabuses per edition, assessment records, faculty qualifications and indicator results shall be retained.
8 Operation
8.1 Operational control
Each edition shall be delivered in accordance with the certified programme. Deviations shall be recorded and justified.
8.2 Conditions for opening an edition
An edition shall not open without assigned faculty meeting 7.2, without available practice data and without published assessment criteria.
8.3 Student assessment
Assessment shall be competence-based, with published criteria and with the possibility of student appeal.
8.4 Post-course follow-up
A requirement specific to this standard. The provider shall contact each cohort at six months to measure real transfer to the job.
9 Performance evaluation
9.1 Monitoring and measurement
The six Annex C indicators shall be measured at the declared frequency, and their evolution between editions analysed.
9.2 Internal audit
The audit programme shall include, at least once per cycle, direct observation of a session being delivered, not only review of the syllabus.
9.3 Management review
The review shall include indicator results, complaints from students and client companies, and the currency of technical content.
10 Improvement
10.1 Continual improvement
The provider shall continually improve the suitability and effectiveness of its programmes, taking measured transfer as the primary signal.
10.2 Nonconformity and corrective action
On a nonconformity the provider shall react, correct it, evaluate the need to eliminate its causes and implement the necessary actions.
Annex A (normative) — Training Programme Requirements
The 16 requirements each programme in scope shall meet. They are verified against the syllabus, the materials and through direct observation of a session being delivered.
| Code | Requirement |
|---|---|
| P.1.1 | AI applied to a real processThe programme shall address AI on specific manufacturing processes, not only general machine learning theory. |
| P.1.2 | Authentic industrial dataExercises shall use real plant data, or synthetic data generated from real processes, with their noise and gaps. |
| P.1.3 | IT/OT convergenceThe programme shall cover how a model connects to the control layer and the limits of that connection. |
| P.1.4 | Full lifecycleThe syllabus shall reach operation, drift and model decommissioning, not end at training. |
| Code | Requirement |
|---|---|
| P.2.1 | Safe architecturesAI/control separation, safe degradation and human override shall be treated as core content. |
| P.2.2 | Historical incidentsReal automation and AI failure cases shall be studied, with their root cause analysis. |
| P.2.3 | Governance and regulationThe AI Act, NIS2 and the applicable sectoral framework shall form part of the programme. |
| P.2.4 | Model limitsStudents shall learn to recognise when a model is outside its validity domain. |
| Code | Requirement |
|---|---|
| P.3.1 | Design of assisted rolesStudents shall be taught to design the job alongside the system, not after it. |
| P.3.2 | Change managementThe programme shall include how an AI system is introduced into a shop floor team without breaking it. |
| P.3.3 | Reskilling strategiesHow the task transition of affected personnel is planned shall be addressed. |
| P.3.4 | Dialogue with worker representationThe framework for informing and consulting worker representatives shall be covered. |
| Code | Requirement |
|---|---|
| P.4.1 | Project in a real settingAn applied piece of work with plant data or scenarios shall be required, assessed against published criteria. |
| P.4.2 | Competence-based assessmentAssessment shall measure what the person can do, not what they can recall. |
| P.4.3 | Published assessment criteriaStudents shall know in advance how they are assessed and against which thresholds. |
| P.4.4 | Post-course follow-upThe provider shall contact students at six months to measure transfer to the job. |
Annex B (normative) — Faculty Requirements
Who may teach. It is the requirement that stops the most applications, and the one that most distinguishes this standard: knowing AI is not enough, you must have been on a shop floor.
| Code | Requirement |
|---|---|
| B.1 | Evidenced shop floor experienceInstructors on technical modules shall evidence 2,000 hours of work in a real industrial environment. Employment history, delivered projects or IRIS-PRO Expert certification or above are accepted. |
| B.2 | Current technical competenceFaculty shall evidence technical or training activity in industrial AI within the last 24 months. |
| B.3 | Safety AI trainingWhoever delivers the P.2 block modules shall evidence specific training in AI safety and governance. |
| B.4 | Independence of judgementWhere the provider is a tool vendor, the solution selection module shall be delivered by someone with no commercial ties to it. |
| B.5 | Supervision ratioThe maximum number of students per instructor in practical sessions shall be declared and respected. |
| B.6 | Update planEach instructor shall have an annual update plan with assigned hours and evidence of completion. |
B.4 exists because a vendor can train excellently on its own product and, at the same time, not be the right party to teach how to choose between products.
Annex C (normative) — Transfer Indicators
The core of the standard. Six indicators with thresholds, measured and declared by the provider and verified in audit. Two consecutive measurements below threshold are a major nonconformity.
| Id | Indicator | What it measures | Threshold |
|---|---|---|---|
| K1 | Application rate at 6 months | Share of trained people who, six months later, have applied what they learned to a real task in their role. | ≥ 60% |
| K2 | Project delivered in a real setting | Share of students completing a piece of work with data or scenarios from a real or highly realistic plant. | ≥ 80% |
| K3 | Safety AI coverage | Share of teaching hours devoted to safe architectures, historical incidents and governance. | ≥ 15% of hours |
| K4 | Faculty shop floor experience | Accumulated hours of real industrial work of the faculty delivering technical modules. | ≥ 2,000 h per technical instructor |
| K5 | Employer satisfaction | Rating of the usefulness of the training by whoever pays for it or receives the trained professional. | ≥ 7 out of 10 |
| K6 | Curriculum currency | Maximum admissible age of technical content without a documented review. | ≤ 12 months |
K1 requires a minimum response rate of 50% of the cohort to be considered valid. A provider without history obtains the certificate with K1 pending and shall supply it at the first surveillance audit.
Annex D (informative) — IRIS-EDU Catalogue
IRIS-EDU 21001 is the certifiable base standard. Codes 21010 to 21050 identify the type of programme in scope, audited over the common baseline.
| Code | Designation | Scope |
|---|---|---|
| IRIS-EDU 21001 | Industrial AI Training — Baseline | Introductory courses, workshops and awareness training |
| IRIS-EDU 21010 | Technical Industrial AI Training | Technical courses on implementation, development and integration |
| IRIS-EDU 21020 | Executive Industrial AI Training | Executive programmes, training for managers and leadership |
| IRIS-EDU 21030 | Master's in Industrial AI | Postgraduate programmes and specialised master's degrees |
| IRIS-EDU 21040 | Safety AI Training | Specific programmes on safety and ethics in industrial AI |
| IRIS-EDU 21050 | Industrial Reskilling Training | Programmes converting shop floor personnel towards AI |
Certification Process
Assessment in three phases. The difference from other management standards lies in phase 2: the audit team sits in on a session and talks to students, rather than only reviewing the syllabus.
Scope, syllabuses, assessment criteria, faculty qualifications and the indicator measurement system.
Attendance at a real session of each programme in scope, review of submitted work and student interviews.
Certificate per named programme, with annual audit and verification of the evolution of the six indicators.
Scheme conditions
- The certificate names the programmes. Adding a new programme to scope requires an audited extension.
- Substantial change to technical syllabus: requires notification and may trigger a new observation.
- K1 pending: a provider without cohorts six months old obtains the certificate with the indicator declared pending.
- Cross-recognition: evidence from a system conforming to ISO 21001 or ISO 29993 is accepted for clauses 4, 5, 7 and 9.
- Relationship with IRIS-PRO: an IRIS-EDU certified programme entitles its students to sit the IRIS-PRO examination at the corresponding level.
Nature of the Scheme and Terms of Use
Transparency about what IRIS-EDU 21001 is and is not
What it is
IRIS-EDU 21001 is a certification scheme owned by European Initiative Industry 4.0, openly published, whose fulfilment is assessed through an independent third-party audit.
What it is not
It is not an ISO, IEC, CEN or UNE standard. It is not accredited by ENAC or any national accreditation body. It does not accredit official qualifications nor replace the competent education authority: an IRIS-EDU certified master's degree still needs its corresponding academic validation.
The numerical coincidence with ISO 21001 indicates equivalence of subject matter, not equivalence of origin or official recognition.
Terms of use of the normative text
The text is published openly and free of charge for reading, citation and internal use, including use as a specification when buying training. Use of the mark and of the designation 'IRIS-EDU 21001 certified programme' is reserved to programmes named on a valid certificate.
Want to Certify Your Programmes to IRIS-EDU 21001?
The usual first step is to compare your syllabuses against the 16 Annex A requirements and check whether your measurement system can sustain the six indicators.
Frequently Asked Questions about IRIS-EDU 21001
Does IRIS-EDU certify my institution or my courses?
Both at once, and neither separately. Clauses 4 to 10 certify the provider's management system; Annex A certifies each specific programme in scope. The certificate names the programmes: one that does not appear is not certified, even if delivered by a certified provider.
We are a manufacturer and we train on our own product. Can we be certified?
Yes. Clause 1.1 expressly includes technology suppliers who train on their products, and in-house training departments. In that case indicator K5 is measured on the customer receiving the training, not on the manufacturer itself.
What exactly is 'transfer' and why is it the axis of the standard?
It is the trained person applying what they learned to a real task in their job. It is measured at six months, through indicator K1, with a 60% threshold. It is the axis because in industrial technical training, classroom satisfaction and shop floor usefulness correlate poorly: a course can be very well liked and change nothing. What industry pays for is the second.
Can fully online training be certified?
Yes, the delivery mode is not in itself an obstacle. What cannot be avoided is requirement P.4.1: applied work with data or scenarios from a real or highly realistic plant. An online programme that solves this with authentic industrial datasets is certifiable; one using only textbook examples is not.
What is required of the faculty?
For technical modules, evidenced experience of work in a real industrial environment: indicator K4 sets 2,000 hours per technical instructor. It is not an academic requirement but a shop floor one. It can be evidenced through employment history, delivered projects or IRIS-PRO certification at Expert level or above.
Is it an ISO standard, or accredited?
No. IRIS-EDU 21001 is a scheme owned by European Initiative Industry 4.0, independent of ISO, IEC, CEN, AENOR and ENAC. It takes the structure of ISO 21001 and mirrors its numbering for consistency of subject matter, just as IRIS-SEC 27001 does with ISO/IEC 27001, but it neither derives from it nor implies official recognition. Nor does it accredit official qualifications.
How long does certification take and how long does it last?
Two to four months to issue. The certificate is valid for three years with an annual surveillance audit. Note: the first full measurement of indicator K1 requires half a year to have passed since a cohort, so a provider without history obtains the certificate with K1 declared pending and shall supply it at the first surveillance audit.