Start with the participants’ actual work
To prepare AI training, identify the participants’ roles and the tasks where they need support. Sales, marketing and operations teams face different problems. The programme should recognise those differences and choose exercises each person can connect with their work, with enough context to understand why a particular approach might help.
A sales team might work on preparing conversations, organising notes or sorting incoming requests. In marketing, the need may involve campaign analysis, reviewing an offer or preparing content variants. Operations teams may want to reduce manual copying between tools and organise information spread across different places.
These examples help start the assessment without assuming every task requires AI. Before the session, ask participants for a concrete difficulty and an example of what they have already tried. Their answers reveal familiarity with the tools and help avoid starting with a demonstration that few can follow.
In my training experience, questions become more useful when there is work to examine. At Laboratório da IA, I review projects in two mentoring sessions each month. Corporate training also needs to consider the shared process, who coordinates it and what the organisation permits people to use.
Choose between an introduction, a workshop and ongoing support
The format should match the expected training outcome. An introductory session helps people understand possibilities and limits; a workshop lets them experiment with a defined problem; ongoing support provides opportunities to review application over time. The choice depends on the starting point and the time available for work after the session.
If management needs to decide where to invest, it makes sense to discuss processes, selection criteria and implementation risks. That conversation may not require a lengthy sequence of buttons and settings. If the team has already chosen a problem and needs to build an initial solution, execution and technical review become more prominent.
A workshop needs different preparation from a talk. Access, equipment, accounts and working materials should be confirmed before the day. If part of the session depends on an integration participants cannot authorise, time ends up being spent on access problems. These details should be settled during preparation.
When the aim is to introduce possibilities to a wider audience, a talk about artificial intelligence may be appropriate. If the challenge requires a team to take responsibility for implementation, start with AI consulting for businesses. Clarifying this distinction helps you choose the kind of support you need.
Build an exercise with a beginning, an outcome and a review
A useful exercise starts with an understandable situation and ends with a result participants can evaluate. Define the input material, expected behaviour and situations requiring a pause or a request for help. Reviewing what happened should be part of the session, with time reserved to discuss difficulties and compare different attempts.
Imagine an exercise preparing a summary of a fictional sales enquiry. The team receives a message naming a service, a deadline and some questions. The assistant should organise that information without inventing a quote or filling in missing data. Participants compare the response with the original request and identify anything lost along the way.
On a second attempt, change the message to include two services or an ambiguous deadline. The challenge becomes recognising when an apparently complete answer hides an incorrect interpretation. That comparison teaches better instructions and, above all, how to decide what needs human review before the process continues.
For software exercises, I use the work shown in Sextas Ímpares #145, about an application built with Claude Code, as a reference. Planning includes rules, data and permissions. The article about planning an application before coding helps participants prepare questions before training and revisit the reasoning afterwards.
Include security and usage criteria from the start
Training should explain which information may be used, which actions need authorisation and how to check AI-generated results. Those rules need to accompany practical exercises. Learning a tool without understanding usage limits leaves people uncertain precisely when they begin applying the knowledge to their actual work and interacting with company information.
Use fictional data or examples prepared for the session. There is no need to copy a customer database to learn how to organise requests or test a classification. When an exercise needs internal information, the organisation should clarify in advance what may be shared and which environment may be used.
If the programme involves applications, participants should understand the difference between hiding a button and protecting an action. Authorisation must be checked on the server, and private data require access rules. In lesson #145 at 23:43, I discuss the security review associated with the application I built.
It is also worth teaching people to communicate uncertainty. When data are missing, the answer should identify the gap and allow investigation to continue. A campaign report should not conceal that its export is incomplete. I develop these precautions on the AI and automation page, with lesson examples and links to demonstrations.
Prepare for application after the training
Training needs space in the calendar for application afterwards. Choose a task, a responsible person and a date to review what happened. Keep the examples and decisions so the team can continue without depending on participants’ memories or the trainer’s permanent availability whenever a question appears during their subsequent work.
In the book I wrote in 2022, I advocated documenting processes and lessons in a playbook. The idea was to let another person continue the work and avoid repeating known mistakes. Today that record can also include AI instructions, examples of acceptable answers and situations requiring review.
In Sextas Ímpares #147, when discussing investment in learning, I returned to the difficulty of accumulating training without applying it. A concrete question helps: what will we be able to do better over the coming weeks? The answer supports reserving time, choosing a trial and returning to the discussion with firsthand experience.
During the review, examine execution and quality. A task may become faster but require more corrections; another may take the same time while making information easier for colleagues to find. Record that difference and decide what to keep. Participants’ experience should inform the next training decision, without making every new release an automatic purchase.
How to request training for your team
To request training, describe who will attend, what they want to learn and the context in which they will apply that knowledge. Include the preferred format, group size and possible dates. My commercial team gathers this information and follows up on the request to discuss conditions and availability with the organisation.
It helps to know whether participants already use AI tools, whether a process has been chosen and whether they want an introduction or practical work. If you are still exploring the topic, say so. We can start by clarifying the objective without pretending a fixed programme already fits a need we have not yet understood.
I have worked in digital since 2012 and have had more than 30,000 students throughout my career. That experience includes marketing, online business and training; current AI demonstrations connect to the work I continue to develop. You can explore my background and the available training before preparing an invitation.
Send the context through the training requests and invitations page. The form captures the need and helps the team organise the next conversation. Dates, programme, duration and budget are discussed afterwards, according to the objective and conditions of the request.