Teach · Managers
How to train your team to use AI
Training a team to use AI means teaching each role to apply AI tools safely and effectively to the work the role already does. Effective AI training starts with an approved tool list and a written policy, uses real tasks instead of generic demos, and continues after the workshop through shared playbooks and coaching.
- Reading time
- 8 min
- Chapters
- 6
- Last reviewed
- November 10, 2026
- Author
- Lingows AI Team
- Difficulty
- Beginner
Before you start
- Approved AI tools, or a decision on which tools to approve
- A manager in each team willing to supply real tasks
- A place to store a shared playbook, such as an internal wiki
The steps at a glance
- Step 01
Survey current AI use
Learn which tools staff already use, for which tasks, and with what data.
- Step 02
Write and share the AI use policy
Approve tools, restrict data, and define when a person must review output.
- Step 03
Plan sessions by role
Design separate sessions for executives, managers, frontline staff, and technical teams.
- Step 04
Train on real work
Use tasks participants bring from the past week and practice live during the session.
- Step 05
Build a shared playbook
Collect prompts that worked, test the prompts on new examples, and name an owner.
- Step 06
Measure and follow up
Track adoption and outcomes, hold office hours, and schedule refreshers.
Chapter 01
How do you find out how a team uses AI today?
Finding out how a team uses AI today takes a short anonymous survey and a few conversations about which tools people already use, for which tasks, and with what data. Many staff already use public AI assistants without formal guidance. Knowing current habits shows where training is urgent and where policy gaps already create risk.
Make the survey anonymous and make clear that the goal is to learn, not to punish. People who have been pasting customer data into a free chatbot will only say so if admitting it is safe. That information is exactly what the policy and the training need to address.
Ask about tasks rather than tools. Questions such as what people use AI for, and what people would like help with, produce better answers than a list of product names. The answers also become the raw material for role-based exercises later in the program.
Chapter 02
What should an AI use policy cover before training starts?
An AI use policy should cover which AI tools are approved, which data may never be entered into an AI tool, when a person must review AI output, how to label AI-assisted work where required, and whom to ask with questions. Training without a policy teaches skills but leaves staff guessing about what the company allows.
AI use policy outline
| Section | What to decide |
|---|---|
| Approved tools | Which assistants and AI features staff may use, and on which accounts |
| Restricted data | Customer, employee, financial, health, and confidential data that must stay out of unapproved tools |
| Human review | Which outputs a person must check before use, such as anything sent to customers |
| Disclosure | When AI-assisted work must be labeled for clients, regulators, or internal records |
| Accountability | The person who signed off remains responsible for the result, regardless of the tool |
| Support | Who answers questions, and how new tools get requested and reviewed |
Keep the policy short and specific. A two-page document that names real tools and real data types is more useful than a long statement of principles. The answer on whether AI is safe for business data covers the data questions most policies need to settle.
Chapter 03
Why should AI training be organized by role?
AI training should be organized by role because executives, managers, frontline staff, and technical teams use AI for different decisions and different tasks. Role-based sessions use examples each group recognizes, so people leave with techniques the group can apply the next working day instead of general awareness that fades quickly.
Training by role
| Role | Focus | Session format |
|---|---|---|
| Executives | Where AI fits in strategy, risk, governance, and budget decisions | Briefing with discussion of the company's own roadmap |
| Managers | Choosing use cases, setting review steps, and measuring adoption | Workshop using processes the managers own |
| Frontline staff | Drafting, summarizing, research, and data cleanup in approved tools | Hands-on practice on real tasks from the team |
| Technical staff | Integrations, evaluation, security, and connecting AI to systems | Technical deep dives and guided builds |
Mixed sessions still have a place for company-wide awareness, but the skill building happens in role groups. A finance team learning to reconcile statements with AI help needs different examples than a sales team drafting proposals.
Chapter 04
How do you make AI training stick?
AI training sticks when every exercise uses real tasks from the participants' own work, when people practice during the session rather than watch a demo, and when follow-up continues for weeks afterward. Office hours, a shared channel for questions, and short refresher sessions turn a one-time workshop into a working habit.
Ask participants to bring two or three tasks from the past week before the session. Work through those tasks live. People remember the moment an assistant cut an hour from their own work far longer than a polished demonstration on someone else's problem.
Teach the basics that transfer across tools: giving clear context, providing examples, asking for a specific format, checking facts against sources, and iterating on a draft. Those habits hold even as the tools change. The prompt engineering answer covers the fundamentals in more detail.
Chapter 05
What is an AI prompt playbook and how do teams build one?
An AI prompt playbook is a shared, maintained collection of prompts, examples, and instructions for tasks a team repeats often. Teams build a playbook by collecting prompts that worked during training, testing the prompts on new examples, naming an owner for upkeep, and storing the playbook where everyone already looks for team documentation.
A playbook entry should state the task, the prompt, an example input, an example of good output, and the checks a person should run before using the result. Short entries that people actually reuse beat long libraries nobody opens. A sample entry looks like this:
Task: Summarize a customer call transcript for the account record.
Context: You are helping an account manager at our company.
Instructions: Summarize the transcript below in five bullet points.
Include: the customer's main request, any commitments we made,
open questions, and next steps with owners.
Do not include: personal details beyond the customer's name and company.
Format: plain bullet points, no more than 120 words.
Transcript: [paste transcript]Review the playbook on a schedule. Retire entries that no longer work well, and update entries when the team adopts a new tool or model.
Chapter 06
How do you measure whether AI training worked?
Measuring whether AI training worked means tracking adoption and outcomes, not attendance. Useful measures include how many people use approved tools each week, which tasks people use AI for, time saved on measured processes, error rates in reviewed output, and the questions staff still ask. Review the measures with managers every month or quarter.
Checklist
AI training program checklist
Lingows AI runs this kind of program in AI training engagements, with sessions by role and playbooks the team keeps afterward.
01Summary
Key takeaways
What to remember
- Survey how the team already uses AI before planning any training.
- Write a short AI use policy before training starts, not after.
- Organize sessions by role and use real tasks from each team.
- Teach verification so staff catch confident but wrong output.
- Keep a shared prompt playbook with a named owner.
- Measure adoption and outcomes, not attendance.
02FAQ
Questions about this guide
How long should AI training take?
AI training length depends on the role and the tools involved. An executive briefing and a hands-on workshop for frontline staff can each fit in a single session. Lasting skill comes from follow-up, so plan office hours, a question channel, and refreshers after the first session rather than one long course.
Do employees need technical skills to learn AI tools?
Employees do not need technical skills to learn everyday AI assistants. Clear writing, good judgment, and knowledge of the work matter more. Technical skills become important for connecting AI to company systems, evaluating output at scale, and building automated workflows, which is why technical staff get separate sessions.
What should staff never put into an AI tool?
Staff should never put restricted data into an unapproved AI tool. Restricted data typically includes customer personal details, employee records, financial account information, health information, passwords, and confidential contracts. The company AI use policy should list restricted data types explicitly and name the approved tools where some sensitive work is allowed.
How do you handle employees who are worried about AI?
Employees who are worried about AI respond best to honesty about intent, a say in which tasks change, and training that shows AI as help with tedious work. Leaders should explain how the company plans to use AI, what will not change, and how staff can raise concerns without penalty.
Should training cover more than one AI model?
Training should cover the approved tools first, then the habits that transfer across models from providers such as OpenAI, Anthropic, Google, and others. Clear context, examples, format instructions, and verification work in every major assistant. Teams that learn transferable habits adapt quickly when the company changes or adds tools.
Written by
Lingows AI Team
Engineering and strategy team, Lingows AI
The Lingows AI team has built AI systems since 2017 and software since 2012. Collective posts cover model landscape changes, tooling selection, and the direction of applied AI.
Put this guide to work with AI Training.
AI training puts this guide into practice with your own teams, on your own work, with a policy and playbooks they keep.