Generative AI fundamentals
Understand what generative AI can do, where it makes mistakes and how to evaluate its responses before using it for specific work tasks. The method is adapted to the tool your team uses.
Find your starting point
A good fit when you need a shared foundation for recognising suitable AI uses and checking responses before applying them at work.
Another option to consider
Strategic AI adoptionIf your main task is prioritising use cases and defining a team pilot, choose strategic adoption.- Hours and sessions
- 6 hours live · 3 sessions of 2 h.
- Group size
- Up to 12 participants per trainer as a reference, adjusted to the practice and feedback.
What you will learn
- 01
Distinguish a suitable AI task from one needing a different approach.
- 02
Explain a response's limits and what information needs checking.
- 03
Apply a basic use and review checklist to a work exercise.
A proposed programme
You will turn a fictional internal guide into a short explanation and a draft email for a new colleague. You will explore how AI helps organise and rephrase information, check its responses against the guide and decide what to use, correct or discard. You will leave with useful material and a checklist for repeating the task with judgement.
We adapt the selection and depth of the modules to the group's starting point and goals.
01Understand generative AI
Generative models, context and types of tasks through accessible examples.
You will work on
- Explain how the model uses instructions and context, and why a convincing answer may be wrong.
- Distinguish synthesis, rephrasing and ideas from tasks needing a source, calculation or decision.
- Define the reader and result before requesting a short explanation of a work guide.
- Practice
- Ask for an initial explanation of the support guide for a new colleague and compare it with the original. Also consider how to count requests in a spreadsheet and confirm an undocumented response time.
- How you will check it
- An initial draft and a reasoned choice of tool or person for each of the three tasks.
02Recognise limits
Errors, incomplete information, variability and criteria for checking responses.
You will work on
- Compare two drafts for clarity, coverage and fidelity to the guide.
- Separate confirmed and missing information from unsupported claims; ask for clarification where needed.
- Decide what to retain, correct or remove, checking omissions and wording.
- Practice
- Review two classroom variants: one keeps the support channel and hours; another adds a guaranteed response within two hours that the guide never states. Prepare a clear, faithful version.
- How you will check it
- A corrected explanation with review decisions justified against the guide, including the response time that remains unconfirmed.
03Start with a use case
Task selection, authorised data and instructions followed by a final review.
You will work on
- Choose an authorised task and material; specify the objective, reader, source and format.
- Turn the explanation into a short email containing the information needed to act.
- Check the email against the guide, review its tone and leave a checking routine.
- Practice
- Prepare a reference card and draft reply explaining how to request support, keeping the documented channel and hours without promising an unknown response time.
- How you will check it
- A reviewed reference card and email, together with the source, instruction and a short checking checklist.
From an internal guide to a clear reply
Teaching example
You will compare AI responses to a sample task and apply a review checklist to decide what to use, correct or discard.
Fictional classroom case. A new colleague asks how to request support. The guide specifies the internal portal and support hours of 09:00–17:00; it gives no response-time commitment. It is supplied as text, with two teaching drafts for comparison.
- Starting point
- The information sits in a lengthy guide. We need a short reference card and email explaining where to request help, what to include and when support is available, using only the information provided.
- First attempt
- One version organises the steps well but is too technical. Another variant prepared for the exercise adds a 'guaranteed response within two hours'. The group keeps the useful structure, clarifies the wording and flags the unsupported promise.
- What we review
- The trainer asks participants to locate the channel and hours in the source and leave the response time unconfirmed. The group then requests an explanation for a new colleague and a short email, checks both and asks the responsible person if a response time is needed.
- Result
- The reference card and email explain how to request support using the correct channel and hours. The group keeps clearer wording, corrects jargon and removes the invented deadline. The draft, guide and review decisions are retained; no email is sent during the exercise.
Information people can understand and use
Original guide
Internal portal
Documented channel for requesting help
09:00–17:00
Support hours · no response time stated
Short explanation
Understandable steps
The same information for a new colleague
Check against the guide
Use, correct or discard each part
Reviewed reply
Reference card and email
Clear channel and hours · no added promises
Response time to confirm
Ask the responsible person if needed
Illustration of a fictional exercise, not a screenshot or client result. Comparison drafts are prepared for the classroom.
Before calling it complete
- The reference card and email are actionable and preserve the guide's internal portal and 09:00–17:00 support hours.
- Clearer wording is kept, errors are corrected and the undocumented response time is discarded or queried.
- The group explains when to use AI for rephrasing, a spreadsheet for a verifiable count and a person to confirm a commitment.
Sessions to put it into practice.
Reference edition · 6 h live
Confirmed reference format: 6 live hours in 3 sessions of 2 hours. Practice takes place during the sessions using fictional texts and is adapted to the client's tool. Features and access are checked; no integrations or specific paid features are required.
- Session 12 h
Choose a task and get an initial draft
Understand response generation, distinguish task types and rephrase the guide for a new colleague.
By the endAn initial draft and an approach for writing, counting and confirming information.
- Session 22 h
Compare responses with their source
Review clarity, facts, omissions and unsupported claims in two classroom variants.
By the endA corrected explanation with decisions on what to use, correct or discard.
- Session 32 h
Deliver a useful explanation and email
Prepare the final material, exchange reviews and document a checklist for another team task.
By the endA reference card and email draft ready for human review, with their source and usage checklist.
If your team needs to go further
- Connecting company documents, deploying an assistant or automating replies requires additional scope and a review of information access.
- Working with real sensitive data and defining company AI policies require specific preparation; this introductory exercise does not address their implementation.
Learn to apply AI to everyday work.

Borja Martínez
The Optimal Flow consultant
Over two years of experience developing projects for companies across industries and sizes.
Borja is the planned trainer for this AI programme. Dates, language and availability are confirmed when arranging the course.
With guidance, practice and support.
Each delivery includes working materials, feedback on practical work and assessment against the agreed objectives.
Materials you can keep
Downloadable, versioned guides, exercises and review criteria for internal use.
To prepare and complete the delivery
A short recommendation, an editable communications kit and a group learning report.
TOF’s own certificate
A certificate of achievement after the agreed assessment. It is not a vendor certification.
A month of email support
On the content covered, up to two accumulated hours for the whole group.
Your team carries out the practice. Implementation services are agreed separately. See the scope →