The secret method to start with ChatGPT, Claude, and Gemini from scratch

The best way to start with ChatGPT, Claude, and Gemini is to test them with everyday tasks and compare their answers. This is the central idea of the masterclass organized by La Vanguardia: to explain from scratch what each AI assistant does (a tool capable of responding to written instructions), when it can be useful, and how to detect its errors. The approach is designed for those who still view these tools from a distance, not for specialists.

What each AI assistant offers

Why does it matter to choose well between ChatGPT, Claude, and Gemini?

It matters because not all doubts or tasks are posed in the same way. La Vanguardia’s session starts from a very practical question: which tool is best to use in each situation and how to ask it for what we need to get a useful answer.

ChatGPT, Claude, and Gemini have become familiar names in technology, but knowing their names does not mean knowing how to work with them. For many users, the difficulty arises as soon as they open them: they don't know what to write, they expect an answer that is too general, or they are not sure if they should accept what they receive without reviewing it.

The masterclass proposes comparing their answers in real time. This detail is especially useful for beginners because it turns an abstract explanation into a visible demonstration: the same request can be used to observe what each assistant answers, how it organizes information, and which aspects should be reviewed.

What tasks can be proposed from day one

The source cites specific uses that are part of many people’s routines. The idea is not to look for an extraordinary application but to start with recognizable and easily verifiable tasks.

• Draft an email with a certain tone.
• Summarize a document to locate its main ideas.
• Plan a trip based on the information you want to organize.
• Prepare a meeting and order the issues to be addressed.
• Learn about a new topic through successive questions.

The editorial reading of this proposal is clear: the first benefit is not in asking spectacular questions but in removing repetitive work and better organizing information. That is often the first surprise. A tool that seemed reserved for experts can be useful for an email or a weekly meeting.

How to learn to ask for better answers

What changes when a request is well formulated?

A well-formulated request allows you to clearly explain what is needed, why it is needed, and what type of response is expected. The masterclass will teach how to formulate good requests and compare the results among the three assistants.

The difference is not about using technical words. It consists of specifying the task. Requesting to draft an email, for example, does not offer the same framework as indicating the subject, desired tone, and purpose of the message. The source does not fix a unique formula but does position the quality of instructions as one of the session’s keys.

To start, the reader can turn any everyday task into an orderly request. First, explain the action; then, add the available context; finally, indicate the expected format. This is a simple way to avoid the assistant having to guess what is wanted.

• Describe the specific task: draft, summarize, plan, prepare, or explain.
• Include the necessary context so that the answer is not disconnected from the objective.
• Indicate the result you seek, such as an email, a summary, a list, or an explanation.
• Compare the answer with that of another assistant when you want to evaluate options.

How to detect errors before using an answer

The session will also teach how to detect errors. This matters because a clear answer is not necessarily a correct answer, and the masterclass program itself includes the limitations of these tools as part of the learning.

Verification must be part of the process, not something that happens after sharing an email, using a summary, or preparing a meeting with doubtful data. The user can read the entire answer, check if it responds to the original request, and ask for a reformulation when something is unclear.

La Vanguardia’s approach is apt for beginners because it does not present artificial intelligence as an automatic solution. It shows it working and, at the same time, teaches maintaining judgment. The tool proposes; the person decides what is useful and what needs to be reviewed.

• The answer addresses exactly the requested task.
• The text maintains the tone and format that were requested.
• The summary retains the ideas of the original document.
• The planning helps to organize the information without replacing personal review.
• Confusing points are reformulated before being used.

The masterclass to start from scratch

Who can attend and what will they find?

The activity is open to all audiences and requires no prior technical knowledge. For readers who still don’t know where to start, this condition removes a common barrier: no need to master the technology before sitting down to try it.

The event is organized by La Vanguardia on Thursday, October 29 at 9:30 a.m. at Factor Barcelona, located at Roger de Llúria, 71. Attendance is free, although prior registration is required. The source indicates the event is designed for those who want to start from scratch.

The session will be led by Aleix Valls, founder and CEO of WeArtificial. He is a mathematician and civil engineer, holds a PhD in computational fluid mechanics, and is a former CEO of Mobile World Capital Barcelona, where he promoted projects such as 4YFN and the mVenturesBcn investment fund. His career connects research, technology, and practical dissemination.

How the session is organized and what to take away

The masterclass is divided into two blocks. The first will explain what ChatGPT, Claude, and Gemini are, how they work, and what their strengths and limitations are. The second, longer, will focus on practical exercises with real-time demonstrations.

During that practical part, attendees will see how to formulate requests, compare the answers from each tool, and detect errors. They will also observe how to incorporate these assistants into daily routines with judgment, instead of using them without a concrete objective.

To get the most out of the day, it is advisable to arrive with a real task in mind: an email to draft, a document to summarize, a meeting to prepare, or a topic to learn. This way, it will be easier to recognize which instructions produce a useful answer and which must be adjusted.

• Think of an everyday task you want to solve.
• Decide what outcome you need before formulating the request.
• Observe how the answer changes when you add context.
• Compare answers without automatically accepting any as correct.
• Note which way of asking for the task seems clearest to you.

The most valuable idea of this proposal is not to pick a permanent winner among three assistants. It is to learn to recognize what is needed, how to ask for it, and when to review the outcome. For someone starting out, that difference can be more useful than memorizing technical features.

ChatGPT, Claude, and Gemini can be incorporated into daily tasks without prior knowledge, as long as the user formulates clear requests, compares when necessary, and reviews the answers. La Vanguardia’s masterclass offers a practical, free, and open starting point for all audiences to lose the fear of trying these tools with judgment.