The knowledge, teaching skills and certification path that turn willing volunteers into trainers who change how people work.
In Part 1 of Training the AI Trainer Series,we saw that artificial intelligence (AI) usage is growing much faster than AI training, and that most of the training people do receive does not prepare them to work on their own. We also saw why a train-the-trainer model is the fastest way to close that gap.
Now comes the harder question. Once you have selected your trainers, what exactly do you teach them, and how do you make sure they can teach others well?
In this article, you will learn:
Why you should start with a competency modelinstead of a slide deck
The seven knowledge domains every AI trainer must master
The teaching skills that matter just as much as AI skills
A simple four-step teaching method that turns sessions into real skills
How to design and certify your own AI Trainer Academy
What belongs in a ready-to-use trainer toolkit
As always, the approach works across sectors, from healthcare, retail and manufacturing to government, education, and banking and financial services.
Start With a Competency Model, Not a Slide Deck
Clarity about the destination is the first gift you can give any learner.
Too many programs begin by building slides. Start instead by defining what people at each level must be able to do.A simple three-level model works well:
AI Aware (all employees): Understands what AI is and is not, knows the approved tools and the basic rules for safe use.
AI Practitioner (everyday users): Uses approved tools confidently on real tasks, writes effective prompts, verifies outputs and follows policy without reminders.
AI Trainer: Performs at Practitioner level or above, can explain why things work, and can coach others through mistakes in real time.
A good test for trainers is simple:
Can they do it?
Can they explain it?
Can they coach someone else to do it?
All three are required.
If you cannot describe what good looks like, no slide deck will get you there.
The Seven Knowledge Domains Every AI Trainer Must Master
A trainer's depth sets the ceiling for every learner they will ever teach.
AI foundations in plain language: What large language models (LLMs) are, why they sometimes produce confident but wrong answers (hallucinations), what a context window is, and the difference between predictive AI, generative AI and agentic AI. Trainers do not need the math. They need clear analogies.
Your approved tools, in depth: The enterprise AI assistant, the AI features embedded in your productivity suite, and any internal knowledge assistants. Trainers must know the features, the limits, and where the data goes when someone presses enter.
Prompting as a discipline: Trainers should teach a structured approach rather than tricks. Public frameworks such as CO-STAR and RISEN, or my own PICOR framework (Purpose, Input, Context, Output, Rules), give learners a repeatable way to ask for what they need. Trainers should also teach iteration, giving examples, and asking the AI to critique its own output.
Verification and critical thinking: How to check facts, numbers, citations and logic, and how to keep a human in the loop for anything that matters. Every trainer should be able to show a live example of the AI being wrong and how they caught it.
Responsible AI, risk and policy: Data classification, confidential and personal information, intellectual property, bias, acceptable use and escalation paths. In regulated sectors such as banking, insurance and healthcare, this also includes record keeping, customer communication rules and model risk expectations.
Role-based use cases: A tested library of use cases for each function. Examples include drafting variance commentary in finance, preparing job descriptions in Human Resources (HR), summarizing account history before a sales meeting, drafting customer service replies, explaining legacy code in engineering, and looking up procedures in operations.
Agentic AI awareness: As AI agents begin to take actions rather than just answer questions, trainers must explain delegation, permissions, approvals and why a human remains accountable for the outcome.
A trainer who knows the tool but not the risk is a fast way to scale bad habits.
Teaching Skills Matter as Much as AI Skills
Knowing AI makes you an expert, but helping others master it makes you a leader.
Many technical experts struggle in the classroom. Give your trainers the teaching skills they need:
Adult learning principles: Adults learn when content is relevant to their work, builds on their experience and solves a real problem.
Facilitating mixed-skill groups: Every room has beginners, power users and skeptics. Trainers need ways to keep all three engaged, in person and online.
Handling fear and resistance: People will ask whether AI will take their job. Trainers need honest, respectful talking points rather than slogans.
Managing live demos: Generative AI does not give the same answer twice. A demo that goes sideways is a gift: it is the perfect moment to teach verification.
Asking good questions: "What would you check before sending this?" teaches more than any slide.
The best AI trainers do not hide the AI's mistakes. They teach with them.
The Teaching Method: Show, Try, Apply, Reflect
Skill is built in the doing, and confidence is built in the reflecting.
Every session your trainers deliver can follow the same four steps:
Show: A short live demonstration on a realistic task, ideally no more than ten minutes.
Try: Learners repeat the task themselves in a guided lab, using an approved tool and safe sample data.
Apply: Learners bring a real task from their own job and complete it with AI during the session.
Reflect: The group debriefs. What worked? What did the AI get wrong? How did you verify it? What will you do differently tomorrow?
Aim for at least half of every session to be hands-on.If learners leave with one real piece of work done faster and better, they will come back to the tool on Monday.
If the learners' hands are not on the keyboard, they are watching AI, not learning it.
Designing Your AI Trainer Academy
Excellence in the classroom is designed long before the first learner walks in.
Here is a sample five-week academy that you can compress or extend to fit your organization:
Week 1, Learner experience: Trainers first take the same course their future learners will take, so they feel it from the learner's side.
Week 2, Depth: Advanced prompting, verification techniques, responsible AI and a policy deep dive, plus an introduction to agentic AI.
Week 3, Use case labs: Each trainer builds and tests three to five use cases from their own function and adds them to the shared library.
Week 4, Facilitation workshop: Teaching skills, handling tough questions, managing virtual sessions and practicing short teach-backs with peer feedback.
Week 5, Certification: A scored teach-back, a practical skills check and a short knowledge assessment.
After certification, each new trainer co-delivers their first session alongside an experienced master trainer before teaching solo.
A simple certification rubric can score:
Accuracy of the content
Share of time that is hands-on
Coverage of responsible AI and verification
Handling of questions and challenges
Time management and clarity
Certify before you deploy. A trainer's first class should never be their first rehearsal.
Equip Trainers with a Ready-to-Use Toolkit
Give trainers great tools and they will spend their energy on people, not on slides.
Do not ask every trainer to build their own materials. Give them a standard kit they can tailor:
A core slide deckwith a facilitator guide and suggested timings
Lab exercises with safe, non-sensitive sample data
A prompt library organized by role
Demo scripts with backup screenshots for when the live demo misbehaves
A Frequently Asked Questions (FAQ) and tough questions guide
A one-page policy quick reference card
Attendance tracking and a short feedback survey
Standard materials protect quality and local examples create relevance, so your toolkit needs both.
Key Takeaways
Build your trainers with care, and they will build an organization that never stops learning.
Define competency levels first, then build content that moves people from one level to the next.
Train trainers across seven domains:foundations, approved tools, prompting, verification, responsible AI, role-based use cases and agentic AI.
Teaching skills deserve as much attention as AI skills.
Use Show, Try, Apply, Reflect in every session, with at least half the time hands-on.
Certify trainers through a teach-back before they go live, and give them a standard toolkit.
Great trainers are not born in a webinar. They are built in an academy.
What is the one thing you wish your AI trainers had been taught before they started teaching? Share your thoughts in the comments.
Previously, Part 1: Who Trains the Trainers? The Hidden Engine of AI Adoption. Next in this series, Part 3: From Classroom to Culture, how to deploy, support and measure your AI trainers.
Explore more: Start Here · Topics · Archive · About