Pāriet uz galveno saturu

Master AI Manufacturing Company Certification · 1. līmenis

AI Foundations for Manufacturing Professionals

Veidojiet atbildīga mākslīgā intelekta pamatu ražošanai, darbam ražošanas zālē, procedūrām, kvalitātei, uzturēšanai, inženierijas atbalstam, piegādes ķēdei, krājumiem, drošībai, darbaspēkam, finansēm, rūpnieciskajām sistēmām, datiem, automatizācijai un ražošanas vadībai.

Līmenis
1
Moduļi
16
Paredzamais laiks
40 stundas
Mācību programma
Versija 2026.4

Sešu līmeņu ceļš

Katrs programmas MAMC līmenis

Līmeņi tiek pabeigti secīgi. Katram ir savi moduļi, zināšanu pārbaudes, noslēguma eksāmens un sertifikāts; 3.–5. līmenī pievienots pārskatīts noslēguma projekts, bet 6. līmenī — atkārtotas mācīšanas (teach-back) uzdevums.

  1. 1. līmenis · Pamati

    AI Foundations for Manufacturing Professionals

    40 stundas · 16 moduļi · Atvērts reģistrācijai

    • Modern Artificial Intelligence in a Manufacturing Company
    • What AI Can and Cannot Reliably Do With Plant Information
    • Export-Controlled Technical Data, Deemed Exports and AI Tools
    • Trade Secrets, Process Knowledge and What Never Leaves the Plant
    • Who Owns the Decision: Engineering, Quality, Safety and Production Accountability
    • Law, Standard, Certification Scheme or Contract: Telling Them Apart
    • un vēl 10
  2. 2. līmenis · Sertificēts profesionālis

    Certified AI Manufacturing Professional

    40 stundas · 14 moduļi · Atvērts reģistrācijai

    • From Foundations to Practice: The Level 2 Operating Model
    • Prompting Architecture for Manufacturing Tasks
    • The Plant Prompt Library: Ownership, Versioning and Retest Triggers
    • Controlled Drafting From the Released Document Set
    • Work Instructions People Can Actually Follow at the Machine
    • Nonconformance, Containment and Corrective Action Drafting
    • un vēl 8
  3. 3. līmenis · Padziļināta darbība

    Advanced AI Manufacturing Operations

    40 stundas · 14 moduļi · Atvērts reģistrācijai

    • From Task to System: The Operations View of Plant AI
    • Mapping a Plant Workflow Before Automating Anything
    • Risk-Tiering Plant AI Tasks by Consequence and Reversibility
    • Writing an SOP a Supervisor Can Enforce on Every Shift
    • Building a Retrieval Corpus From Controlled Plant Documents
    • Retrieval Failure Modes: Superseded Revisions, Wrong Part, Wrong Customer
    • un vēl 8
  4. 4. līmenis · Meistara sertifikācija

    Master AI Manufacturing Company Certification

    40 stundas · 14 moduļi · Atvērts reģistrācijai

    • What a Manufacturer Owes: Accountability for AI Across the Company
    • Writing the Company AI Governance Programme
    • The AI Decision Record and the Evidence Pack
    • Mapping Every Obligation That Binds the Plant
    • Export Control and Controlled Information as a Governance Function
    • Defence Flowdowns, the Certification Programme and Cloud AI Services
    • un vēl 8
  5. 5. līmenis · Automatizācijas speciālists

    AI Manufacturing Automation Specialist

    40 stundas · 14 moduļi · Atvērts reģistrācijai

    • What an Agent May Do in a Plant: Authority and Its Limits
    • The Agent Authority Register and What Never Goes to an Agent
    • Orchestrating a Plant Workflow Step by Step
    • The Integration Map and Data Contracts
    • Read-First Integration With ERP, MES, QMS and CMMS
    • The OT Boundary Nothing Crosses
    • un vēl 8
  6. 6. līmenis · Sertificēts pasniedzējs

    Certified AI Manufacturing Instructor

    40 stundas · 14 moduļi · Atvērts reģistrācijai

    • Teaching Experienced Plant People: Who Is in the Room
    • Designing a Session Backwards From the Behaviour
    • Curriculum Accuracy When Standards, Rules and Tools Change
    • Facilitating Sceptics, and Running the Same Session on Every Shift
    • Running a Demonstration Built Around a Deliberate Failure
    • Designing Hands-On Labs on Fictional Plants and Data
    • un vēl 8

Kursa plāns

1. līmenis detalizēti: ko aptver mācību programma

Katrs modulis savieno praktiskās mākslīgā intelekta spējas ar darbinieku drošību, kvalificētu pārskatīšanu, intelektuālā īpašuma aizsardzību, kiberdrošību, pilnvarotu sistēmu piekļuvi, apstiprinātām procedūrām, ražotāju norādījumiem, auditējamību, eskalāciju un atbildīgiem lēmumiem.

01. modulis

4 nodarbības

Modern Artificial Intelligence in a Manufacturing Company

What the systems sold to a plant as "AI" actually are, and why that changes how you check them. This module separates fixed automation, rules, statistical process control, machine learning and generative AI, explains how a language model builds an answer, opens up the retrieval, tools and agents inside plant products, and maps the categories a small manufacturer meets.

  • Fixed Automation, Rules, Machine Learning and Generative AI 35 min.
  • How a Language Model Builds an Answer 40 min.
  • Retrieval, Tools and Agents Inside the Products You Are Sold 40 min.
  • A Field Guide to Plant AI Categories and the Public Help You Can Reach 45 min.

02. modulis

4 nodarbības

What AI Can and Cannot Reliably Do With Plant Information

Fluent output is not checked output. This module shows how a tool invents a standard, a revision or a part number, how tolerance and unit arithmetic fails silently, how to read an accuracy figure next to the recall it is hiding, and why the strongest countermeasure to over-trust is a habit rather than a warning label.

  • Fabricated Standards, Revisions and References 35 min.
  • Tolerances, Units and the Limits of Context 40 min.
  • Reading Accuracy, Recall and F1 Honestly 45 min.
  • Automation Bias and the Discipline of Doubt 40 min.

03. modulis

4 nodarbības

Export-Controlled Technical Data, Deemed Exports and AI Tools

The sharpest AI trap in manufacturing. This module explains what technical data is, why releasing it to a foreign person inside the United States can itself be an export, what the published-information and end-to-end-encryption provisions really cover, who must register, and how to decide whether a drawing may enter a system at all.

  • What Counts as Controlled Technical Data 40 min.
  • The Deemed Export Rule and What a Tool Does With a File 45 min.
  • Published Information and Encryption: What They Do Not Cover 40 min.
  • Screening a File Before It Reaches Any System 45 min.

04. modulis

4 nodarbības

Trade Secrets, Process Knowledge and What Never Leaves the Plant

Your parameters, your heat-treat recipe, your yields and your cost model are the plant's crown jewels, and federal law protects them only while you keep them secret. This module covers the statutory definition, the reasonable-measures element a pasted prompt can quietly weaken, what a federal civil action involves, the notice employers owe, and how to classify plant data.

  • What Makes Process Knowledge a Trade Secret 35 min.
  • Reasonable Measures and the Outside Tool 45 min.
  • When a Secret Walks: Misappropriation, Remedies and the Notice Employers Owe 40 min.
  • Classifying Plant Data Before a Tool Sees It 40 min.

05. modulis

4 nodarbības

Who Owns the Decision: Engineering, Quality, Safety and Production Accountability

A tool can draft, extract, summarise and flag. It never dispositions a part, releases a lot, seals a drawing or clears a machine. This module covers the duties that do not move when AI helps: the employer's duty under the general duty clause, the named quality authority, the state question of whether a document needs a seal, and a code that binds only its members.

  • The Employer Duty That Does Not Move 35 min.
  • The Named Quality Authority and What AI Never Signs 40 min.
  • Does This Engineering Document Need a Seal? Two State Examples 40 min.
  • A Professional Code Binds Its Members, Not the Plant 35 min.

06. modulis

4 nodarbības

Law, Standard, Certification Scheme or Contract: Telling Them Apart

An OSHA rule is law. A quality standard is a document you buy. Certification is a scheme sold by a registrar or demanded by a customer. And an automotive or aerospace requirement binds you through a purchase agreement, not a statute. This module teaches the four apart, places the quality standards accurately, and shows where the voluntary AI management standard fits.

  • Four Kinds of Obligation in a Plant 40 min.
  • Where the Quality Standards Actually Sit 40 min.
  • A Customer Requirement Is a Contract, Not a Law 40 min.
  • The AI Management Standard and What a Vendor Claim Is Worth 35 min.

07. modulis

4 nodarbības

OSHA Duties and the AI Standard That Does Not Exist

There is no OSHA artificial intelligence standard, directive or guidance document. This module teaches the duties that do bind a manufacturing employer — the general duty clause, machine guarding, the control of hazardous energy, personal protective equipment, hazard communication and injury recordkeeping — and shows how an AI-related hazard is reached through them.

  • The Employer Duty and the OSHA AI Standard That Does Not Exist 35 min.
  • Machine Guarding and Hazardous Energy When Equipment Gets Smarter 40 min.
  • PPE Hazard Assessments and Chemical Information Drafts 35 min.
  • Injury Recordkeeping, State Plans and What National Figures Can Say 35 min.

08. modulis

4 nodarbības

Robots, Cobots and Machine Safety When Perception Is Learned

What changes when a stop, a speed limit or a separation distance depends on a model that can be wrong. This module separates a safety function from a perception feature, works through the robot-system failure modes OSHA itself documents, names the robot and machinery safety standards with their status and edition, and explains why a vision system is never the safeguard.

  • A Safety Function Is Not a Perception Feature 40 min.
  • How Robot Systems Actually Hurt People 40 min.
  • Robot Safety Standards, and Which Edition Applies 40 min.
  • Collaborative Applications, Exoskeletons and the Integration Contract 40 min.

09. modulis

4 nodarbības

Prompt Engineering for Manufacturing Professionals

How to ask for plant work in a way you can check: the data screen that happens before you type, a nine-part prompt built around a supplied released document rather than a recalled one, disciplined iteration, and patterns for work instructions, nonconformance narratives, maintenance summaries and supplier email. Every output is a draft until a named person owns it.

  • The Screen That Happens Before the Prompt 35 min.
  • A Structured Prompt for Plant Work 40 min.
  • Iterating Without Drifting From the Source 35 min.
  • Prompt Patterns for Plant Documents and Supplier Correspondence 40 min.

10. modulis

4 nodarbības

Verifying a Standard, a Regulation and a Revision Level

An assistant that names a standard, a regulation or an edition is giving you a lead to check, not a citation to quote. This module builds a research workflow that confirms a reference exists, is current, and says what is claimed for this product and this customer, using a withdrawn quality standard edition and a superseded medical-device provision as the worked examples.

  • The Assistant Is Not the Authority 35 min.
  • Withdrawn, Superseded or In Revision: Reading a Status Field 40 min.
  • Does This Requirement Reach My Part, My Plant and My State? 40 min.
  • The Verification Record That Survives an Audit 35 min.

11. modulis

4 nodarbības

Drafting Work Instructions, Quality Records and Supplier Communication

How to get a first draft without a first mistake. This module covers drafting work instructions, maintenance procedures, nonconformance narratives and supplier correspondence from a verified set of released documents, what the record must show about preparer, reviewer and evidence, and why no AI tool drafts a disposition, a release or a conformance statement.

  • First Draft Without First Mistake: Working From a Verified Set 40 min.
  • Where Drafting Stops: Safety Content in Instructions and Procedures 40 min.
  • Nonconformance Narratives and Corrective Action: Drafted, Never Dispositioned 40 min.
  • Electronic Records, Audit Trails and What You Send a Customer 40 min.

12. modulis

4 nodarbības

Machine Vision Inspection and Predictive Maintenance: What the Evidence Actually Shows

Read a detection claim like an engineer. This module takes apart accuracy, precision, recall, F1 and mean average precision, shows what the most-used public datasets actually contain, converts a published miss rate into escaped defects on your own volume, and sets out what a pilot on your own material has to look like before anyone believes a number.

  • Accuracy, Precision, Recall, F1 and mAP: What Each One Hides 45 min.
  • What the Public Datasets Actually Are 40 min.
  • From a Miss Rate to Escaped Defects on Your Own Volume 45 min.
  • A Pilot on Your Own Material, and the Trust Problem Behind It 45 min.

13. modulis

4 nodarbības

Building Repeatable Plant AI Workflows

How a one-off chat becomes a controlled plant process: a six-stage workflow canvas with a trigger, a data-class screen, a named AI step, verification, a named approval and a record. It also covers designing so a wrong output cannot become a wrong action, where an agent's authority must stop, and how to keep the review step honest.

  • From a Chat Window to a Controlled Plant Workflow 40 min.
  • Designing So a Wrong Output Cannot Become a Wrong Action 40 min.
  • Agents in the Plant and Where Their Authority Stops 40 min.
  • Keeping the Review Step Honest and Recording the Approval 40 min.

14. modulis

4 nodarbības

Plant Cybersecurity, Vendor Claims and the OT Boundary

Where an AI tool may sit in a plant and where it may not. This module draws the operational-technology boundary, sets the asset inventory and access basics a plant without security staff can run, teaches you to read a vendor's data-handling claims for what they avoid saying, and shows how a federal contract clause turns a recommendation into an obligation.

  • The OT Boundary and Why No AI Tool in This Level Crosses It 40 min.
  • Asset Inventory and Access Basics Without a Security Team 40 min.
  • Reading a Vendor's Security and Data-Handling Claims 45 min.
  • When a Federal Contract Clause Sets Your Security Floor 40 min.

15. modulis

4 nodarbības

Practical Manufacturing AI Lab

Four hands-on labs on fictional plants. Find the planted errors in an AI-drafted work instruction and nonconformance report; catch a fabricated standard citation and a wrong drawing revision, then rebuild the weak prompt; screen a quote package for controlled technical data and trade secrets; and take apart a vision-inspection claim before drafting a plant AI use policy.

  • Lab One: The AI-Drafted Work Instruction and Nonconformance Report 45 min.
  • Lab Two: The Fabricated Citation and the Wrong Revision 45 min.
  • Lab Three: Screening a Quote Package and Designing the Workflow 45 min.
  • Lab Four: The Vendor Claim and the Plant AI Use Policy 45 min.

16. modulis

4 nodarbības

Capstone: The Plant AI Implementation Plan

The plan you will actually write: choose and scope one low-risk problem, classify every data class it touches and set the controls, name the duties that do not move to a tool, ask the vendor the questions that matter, and define the baseline your own plant measures, the failure indicators, the stop conditions and the improvement loop.

  • Choosing and Scoping a Low-Risk First Problem 40 min.
  • The Data Classes in Your Plan and the Controls They Require 40 min.
  • Duties That Do Not Move, the Vendor Question and the Documentation 40 min.
  • The Baseline, the Stop Conditions and the Improvement Loop 45 min.

Mācīšanās rezultāti

Ko jūs varēsiet darīt

  • Novērtējiet mākslīgā intelekta rīkus ražošanas lietojumiem, ierobežojumiem, darbinieku drošībai, datu drošībai, intelektuālajam īpašumam, rūpniecisko sistēmu riskam un profesionālās pārskatīšanas robežām.
  • Veidojiet uzticamas uzvednes un pārbaudes procedūras ražošanai, kvalitātei, uzturēšanai, piegādes ķēdei, krājumiem, apmācībai, pārskatiem un saziņai ar klientiem.
  • Uzlabojiet darbības dokumentāciju un analīzi, nedelegējot iekārtu vadību, drošības, inženierijas, uzturēšanas, kvalitātes, kiberdrošības vai regulatīvos lēmumus modelim.
  • Automatizējiet atkārtotas biroja un rūpnīcas darbplūsmas ar definētu datu piekļuvi, atļaujām, apstiprinājumiem, izņēmumiem, audita pēdām, izvades pārbaudēm un eskalācijas ceļiem.
  • Plānojiet drošas integrācijas un šauri norobežotus aģentus ap pilnvarotām ražošanas sistēmām, uzticamiem darbības datiem, apstiprinātām procedūrām un atbildīgiem īpašniekiem.
  • Izveidojiet ražošanas mākslīgā intelekta ieviešanas ceļvedi ar pārvaldību, darbaspēka attīstību, kiberdrošības kontroli, darbības rādītājiem, ieviešanas prioritātēm un izmērāmu ieguldījumu atdevi.

Novērtēšana

Level 1 Final Comprehensive Examination

Noslēguma novērtējums mēra, cik labi pielietojat mācību programmu reālistiskos ražošanas, kvalitātes, uzturēšanas, piegādes ķēdes, darbaspēka, drošības dokumentācijas, rūpniecisko datu, kiberdrošības, finanšu un vadības scenārijos.

Jautājumi
50
Laika ierobežojums
90 minūtes
Izturēšanas vērtējums
80%
Maksimālais mēģinājumu skaits
3

Vadiet atbildīgu mākslīgā intelekta ieviešanu ražošanā

Sāciet ar pārdevējiem nepiesaistītajiem pamatiem, drošības robežām, pārbaudes kontroli un ražošanas darbplūsmām, kas aplūkotas visā 1. līmeņa mācību programmā.

Neatkarīga profesionālā izglītība

MAMC ir neatkarīga profesionālās izglītības programma. Tā nav inženiera licence, drošības sertifikācija, arodlicence, regulatīvs apstiprinājums vai iekārtu kvalifikācija un neaizstāj kvalificētu inženierijas, uzturēšanas, kvalitātes, drošības, kiberdrošības, juridisko, grāmatvedības vai regulatīvo pārskatīšanu. Mākslīgā intelekta radītie rezultāti jāpārbauda cilvēkam. Mākslīgais intelekts nedrīkst patstāvīgi vadīt rūpnieciskās iekārtas, apiet apstiprinātas drošības sistēmas, atļaut nedrošu iekārtu darbību vai nodrošināt neatļautu piekļuvi rūpnieciskajām sistēmām. Iekārtu darbībai jāatbilst apstiprinātām procedūrām, ražotāju norādījumiem, piemērojamām prasībām un objektam specifiskai kontrolei. Programmu neapstiprina OSHA, NIST, ISO, ASME, IEEE, NFPA, neviena inženieru padome, ražošanas asociācija, arodorganizācija, akreditācijas organizācija, regulators vai valdības aģentūra. Netiek apgalvotas nekādas profesionālās izglītības stundas vai apstiprinājumi, un netiek garantēts neviens ražīguma, kvalitātes, drošības, darbspējas laika, atbilstības, rentabilitātes vai ieguldījumu atdeves rezultāts.