Communicating with an AI is much simpler than it seems, but you need a few keys to get the best possible results. With this guide, you'll know exactly how to phrase your requests so that TOWeb 14's AI can help you effectively in your everyday tasks of creating and updating your website content.
The TOWeb 14 AI agent can be used on any existing paragraph in your site via the (✨) button shown when you hover over it, as well as in the window for adding a new paragraph. Depending on your needs, you can ask the AI agent to generate either an image or text content.
Note: a request sent to an AI may also be called a “query” or a “prompt”.
The golden rule: speak naturally
You don't need to learn a complicated language or use technical words: just write what you want as if you were talking to someone. The AI understands French (as well as many other languages), full sentences and even polite phrases. Don't worry about writing perfect sentences: what matters most is being clear.
A practical example:
✓ Good: "Hello, could you help me understand how a microwave oven works?"
✗ Less good: "Microwave functioning" (too short and vague)
Write as if you were sending an email to a colleague who really wants to help you. Be polite and precise, and don't hesitate to use full sentences.
Precision: your best ally
The second fundamental point is that the more relevant details you give, the better the answer will be. The AI cannot guess what you have in mind; it works only with the information you provide.
Questions to ask yourself before writing:
✓ What exactly is my goal?
✓ Who is concerned? (me, my family, my business...)
✓ Are there any constraints? (budget, time, level...)
✓ What result do I expect?
Comparing vague requests with precise ones
| ✗ Request too vague | ✓ Precise and complete request |
|---|---|
| "Cake recipe" | "I'm looking for an easy, quick chocolate cake recipe (maximum 1 hour of preparation) for my 8-year-old son's birthday. There will be 10 of us." |
| "Help me with my CV" | "I'm applying for a sales manager position in an SME. Can you review my CV and suggest improvements to better highlight my 5 years of sales experience?" |
| "Explain the computer to me" | "I want to understand how a computer works in simple terms. Can you explain the main components (processor, memory, hard drive) as if I were 12 years old?" |
Phrasing your request as precisely as possible gives the AI everything it needs to provide a truly useful answer suited to your situation. This applies not only to text but also to generated images: whether you want the result to look like a realistic photo or rather like a drawing in a particular painting style, whether certain details should appear in the background or not, whether the image should be in warm or pastel tones or in specific colours, etc.
Context: why it matters
Giving context means explaining
- Why you are asking this question
- In what setting (personal, professional, school...)
- Your level of knowledge of the subject (beginner, intermediate...)
- Any constraints (budget, time, equipment available...)
Examples with and without context:
- Without context: "How do I learn English?"
- With context: "I need to prepare for a business trip to England in 3 months. I'm a beginner in English and I have 30 minutes a day to practise. What are the best methods to make quick progress in conversational English?"
Applying a request to your existing text
If your site already has content, you can ask the AI agent to rework it however you like, for example to add or expand on points you may have left out, rephrase a paragraph with less technical vocabulary, in a very friendly/playful or more commercial/sales-oriented tone, translate the paragraph into another language, etc.
You can also ask the AI to structure its answer in a particular way to get exactly what you need.
Examples of formats you can ask for:
- A numbered or bulleted list (for steps or tips):
"Give me 5 tips as a numbered list" - A short summary (in X sentences or Y words)
"Summarise this concept for me in 3 sentences maximum" - A detailed explanation
"Explain this to me very simply, without technical jargon and with examples" - A simplified version
"Explain it to me as if I were 10 years old"
💡 Practical tips for TOWeb 14
- When you choose to apply an AI request to the content of a paragraph, TOWeb's AI agent will automatically add your paragraph's content to the end of your request.
- So that the AI does not mistake part of your request for the content, try as far as possible to refer explicitly to the content that TOWeb will attach. Example of a final sentence to use: "The text content I'd like you to apply my request to is the following:" (with the colon ":" at the end, to refer to what TOWeb will pass as an attachment to the AI agent)
- If you want the bold text, line breaks, bulleted lists and other formatting in your original text to be kept, be sure to say so.
Example: "Correct the spelling of my text while keeping its formatting :"
- Since the AI's first answer may sometimes not satisfy you, we recommend creating a new paragraph or duplicating your existing paragraph. That way you keep your original content (to reuse it, reread it, compare, or simply not lose it) while also asking the AI to modify or rephrase its answer until you get the result you want.
- Before submitting a new AI request, ask yourself whether you might need it again elsewhere or later. If so, remember to tick the option to save your request among the predefined ones so you can reuse it.
In summary
When the AI request you plan to make applies to the text of an existing paragraph in your site, duplicate that paragraph first. This lets you reuse your original paragraph if needed before the AI replaces its content.
Then, before sending your request to TOWeb's AI, check that you have followed these points:
✓ I wrote a complete, natural sentence
✓ I stated my exact goal
✓ I gave the necessary context
✓ I mentioned any constraints
✓ I specified the answer format I want (if needed)
Version 14.02 (or later) is recommended to make full use of all the latest AI capabilities of TOWeb 14, in particular image generation and the use of predefined requests.