Today Asere taught a class.
The session worked because it took a big idea — using AI better — and turned it into something the group could use right away.
Before the training, I prepared an infographic. That was the useful first move. Many training sessions drown while trying to explain everything. This one needed a simple map. Something people could look at and think: “OK, I can start here.”
The broader topic was artificial intelligence, but the part I want to write about today is prompting.
Prompting gives AI direction
Many people start using AI like Google with steroids:
Summarize this.
Or:
Help me with this.
Sometimes the result is decent. Often it sounds fine and does very little. The instruction was weak, so the answer came back weak.
In the training, we landed on a simple formula:
Role + task + context + format.
It looks small, but it changes the conversation.
A weak prompt says:
Make me a proposal.
A prompt with direction says:
Act as an Odoo implementation consultant. Prepare a proposal for a small vehicle rental company that wants to improve invoicing and reservation tracking. Use a clear tone, without unnecessary technical language. Deliver it in sections: scope, phases, risks, and next steps.
The second version has direction.
You tell the AI what role to think from, what work to do, what context it has, and how you want the answer delivered. In practice, that is prompting: asking with more precision.
Context decides the quality
AI works with what you give it. If you hand it two loose sentences, it fills the rest with average.
And average is dangerous because it usually sounds correct.
That is why we talked about organizing context. Use ChatGPT projects when continuity matters. Split conversations by topic. Avoid endless chats where old decisions, changes of direction, and noise get mixed together.
A chat gets dirty too.
Context helps. The trick is giving AI what it needs to work well, instead of piling up information without order.
Speed and judgment are different things
We also talked about speed.
Some tasks only need a quick answer: an initial list, a title, a rewrite, a rough first draft.
Other tasks need more reasoning. Compare options. Check assumptions. Find risks. Say, “I would not choose this option because...”
Fast mode helps you start. Thinking mode helps when you need judgment.
The key is knowing which one you are asking for.
Put one AI to review another AI
One of the strongest ideas in the session was separating roles.
Instead of asking one AI to do everything, you can set up a simple loop:
- One AI proposes.
- Another AI reviews.
The first one writes, designs, or structures. The second one criticizes, finds gaps, spots vague parts, and asks for improvements.
Then you iterate.
That changes the work. You move from copying the first answer to running a process. A small process, but still a process.
That is one of the useful lessons: using AI well means designing the workflow better.
Boundaries are part of the prompt
A good prompt says what you want and where the edges are.
You can ask it to use only the information you gave it. To mark doubts. To avoid a sales tone. To return the answer as a table, bullet points, or steps. To say when it is guessing.
That cuts down the nonsense.
And when you are training other people, that matters. You are teaching a way to think with the tool.
The infographic did its job
The infographic reduced friction.
It put a structure in front of the group. Everyone could return to it. The conversation moved faster because there was a shared reference point.
That is a very real use of AI in training: helping the trainer prepare better material in less time.
The session could have drifted into the usual big phrase: “the future of artificial intelligence.” Instead, it landed on the ground:
if you want better answers, learn to give better instructions.
Professor Asere
I like the name “Professor Asere” because it has a joke inside it, and also some truth.
Asere organized the content, summarized it, turned it into an image, and helped turn a wide explanation into something teachable.
That is the point.
AI works better as a teacher when someone directs it well. It is useful as a tool for thinking, summarizing, contrasting, and explaining.
And directing it starts with the prompt.
The perfect prompt is a fantasy. The useful part starts earlier: tell the AI who it should be, what it should do, what context it has, and how you want the result.
Role. Task. Context. Format.
Four words. Today they were enough to help finish a full training session in record time.