For a long time, I avoided putting artificial intelligence front and centre when presenting my projects. The term could make people uneasy and take over the conversation before we had even discussed their needs.
Today, those conversations have changed. Many people tell me they already use AI. They have tried an assistant, ask it questions, draft text or look for help with a task.
Yet when we look together at how they use these tools, I often see the same pattern: everyone is working with whatever they have figured out themselves. They try a wording, ask again and keep an answer that seems suitable. It can be helpful, but the results are inconsistent.
I am training more and more people to make better use of these tools. What stands out is the importance of taking time to understand how they work: what information to provide, how to explain a need and how to check the response.
That is also the purpose of my work with businesses in and around La Rochelle: helping people develop a practice they can return to in their everyday work.
A simple interface can leave a lot of questions unanswered
A chat window makes everything look straightforward. You type a question and an answer appears.
For someone new to the tool, the important questions come afterwards. Do I need to explain my role in every conversation? Does the assistant already know the company? Can it access the documents I mention? Why does a request work well one day and less well the next?
Regular AI users gradually develop certain habits. They explain the context, provide an example and distinguish available information from missing details. Those habits are not obvious when you are starting out.
Training makes them easier to understand. It also gives people space to ask questions they might hesitate to raise when everyone else seems to know how to use the tool already.
This matters: someone can use an assistant regularly without understanding what affects the quality of its responses.
Context is everything AI cannot work out on its own
Within a company, colleagues do not need to explain everything to one another.
They know the customers, working habits, industry terminology and commitments they can make. They know which document is the reference and who needs to approve something.
An assistant does not automatically have that knowledge.
Take a simple example: preparing a reply to a customer enquiry. The situation may seem clear to us because we have followed the earlier exchanges. The tool may be missing that history, the purpose of the reply, a deadline or a sales policy.
Providing good context means making those details available and understandable.
It does not mean dropping every company document into a conversation. An outdated reference, contradictory instructions or large amounts of irrelevant information can make the task harder.
Useful context provides the right elements: the need, current information, the rules to follow and, where helpful, an example of the expected result.
Learning to explain your work is part of the training
When preparing a request for AI, we sometimes have to spell out things we normally do without thinking.
What should these meeting notes contain? What makes this response satisfactory? Which details must be kept? When should we ask for clarification rather than fill in missing information?
Working through these questions helps us understand what we actually expect from the tool.
For example, a request for a summary becomes more precise when we know who will read it and which decision it should support. A summary for preparing a meeting may serve a different purpose from a document sent to a colleague who is new to the case.
Taking time during training to explain these differences helps people move beyond trial and error. They learn to prepare their requests, understand why a response falls short and make more focused corrections.
They also learn to keep a critical eye. A well-written answer can contain a mistake. Precise context improves the conditions for doing the work, but it does not remove the need to check it.
A shared professional workspace helps the whole team
Progress remains fragile when everyone keeps their instructions, documents and useful habits to themselves.
One person uses an old presentation. Another has built a helpful template nobody else knows about. A new colleague repeats the same experiments because they do not know what is already available.
A shared professional workspace gives the team common reference points:
- the tools they use and their intended purposes;
- up-to-date reference documents;
- reusable instructions and examples;
- the information each person may access or share;
- the outputs that require checking or approval.
This does not mean sharing one account or giving everyone access to every folder. Each person should retain access appropriate to their role. The aim is to provide consistent references to the people who need them.
Someone also needs to keep those references current. An instruction that is useful today may become wrong after a change to an offering, a team structure or a procedure.
Training then becomes a team effort: it helps people build and maintain this shared environment.
A few shared standards make independent work easier
The word “standardise” can sound rigid. In practice, it may simply mean keeping a structure for meeting notes, specifying the information a request needs or sharing the criteria for an acceptable response.
These reference points save people from starting from scratch every time.
They also leave room for judgement. A common template helps prepare the work; the person using it can still adapt it to a particular situation.
This is the change that interests me in training: seeing AI use become more deliberate, more consistent and easier to return to day after day. People better understand what they can ask for, what is missing and what they still need to check.
The benefits depend on the tasks and the people involved. To assess them, we need to look at the whole job: preparation time, necessary corrections and the quality of the usable result.
What the team clarifies today will also help AI agents
This way of working has longer-term value too.
Agents that can work within tools and carry out several steps of a task are developing. Their role in companies will vary with business needs, resources and the level of control people want to retain.
As we give them more actions to perform, the quality of their context will become even more important.
An agent preparing a customer follow-up will need to know which file to consult, which information is authoritative, which rules apply and when to involve a person. Access to tools alone will not give it that understanding.
To me, training teams today is part of that preparation. Clear documentation, explicit responsibilities and checking habits can provide a foundation for the agents a company chooses to introduce alongside its staff.
Their access will still need to be defined, their behaviour tested and their actions monitored. But a team that already knows how to explain its work will have a stronger starting point for assigning them a task.
AI training means building habits that last
The tools will keep evolving. Interfaces, features and the range of actions they support will continue to change.
Knowing how to express a need, choose a reliable source, provide useful context and evaluate a result will remain valuable.
That is the independence I aim to develop through training: helping people use today's tools more effectively and understand how to approach the next ones.
I support businesses in and around La Rochelle with this work, starting from their day-to-day tasks, their documents and their teams' questions. You can find this approach among my AI services for businesses in La Rochelle (in French).
Does your team already use AI, with everyone following their own habits? Let’s build shared reference points that fit the way you work.