Skip to content
Let’s plan the right software for your processes. Call us for a demo or a quote: +90 546 737 48 29

TR EN DE

How to use AI

AI chat assistants such as ChatGPT, Claude, Gemini, Microsoft Copilot and Perplexity are tools that reply in natural language to the question or instruction you type. They save time on tasks such as writing text, summarising, generating ideas, preparing tables and getting to grips with a topic. The key to getting good results from an assistant is to explain clearly what you want. The instruction you write to the assistant is called a prompt.

In brief

  • For good results, give context, state the task and the format clearly, and add an example.
  • Refine the first answer step by step with follow-up instructions.
  • AI can produce incorrect information; verify anything important.
  • Do not enter personal data or confidential company data.

Note

This field changes quickly: the features and terms of use of the tools are updated often. This guide gives general information as of October 2026; for details, see the official page of the tool you use.

On this page

Steps to writing a good prompt

  1. Give context

    The assistant does not know you, your work or your aim. Explain in a sentence or two who you are, who the text is for and what it will be used for. Instead of "Write an email", saying "I am a sales representative at a company that manufactures machine parts; I need to write an apology email to a customer whose delivery is delayed" gives a far more accurate result.

    Diagram of a prompt's structure: context, task, format, example and limits : Enlarge

  2. State the task and the format clearly

    Say what you want and what the result should look like: length, tone (formal or friendly), a bulleted list, a table or headings. Instead of "Keep it short", saying "No more than 120 words, in three paragraphs" gives a better result.

    Comparison: a vague prompt and a strong prompt that includes context, task and format : Enlarge

  3. Give an example

    A text you like, a template you use or a small sample of the output you want makes it easier for the assistant to pick up the style and structure. If the example contains personal or confidential information, remove it first.

    Chat illustration: the user provides a sample text and asks for a new text in the same style, and the assistant replies : Enlarge

  4. Work step by step and refine

    The first answer is rarely perfect. Improve the result with follow-up instructions such as "Make it shorter", "Rewrite the second paragraph in a more formal tone" or "Add a price column to the table". Break complex jobs into parts: ask for an outline first, then have each section written separately.

    Chat illustration: the user shortens and corrects the first answer with follow-up instructions : Enlarge

Example prompts

  • Writing an email: "Write a short, polite email to our customer telling them that their order will be delayed by one week, apologising and giving 15 November as the new delivery date. Use a formal but warm tone and do not exceed 120 words."
  • Summarising: "Summarise the meeting notes below in 5 bullet points. List the decisions made and the people responsible separately: [paste the notes here]"
  • Extracting a table: "Turn the products mentioned in the text below into a table with the columns product name, quantity and unit price. Write 'not specified' in any row where no price is given."
  • Understanding a topic: "Explain the difference between IMAP and POP3 as you would to an employee with no technical background, using an analogy from everyday life."

Limitations: check every answer

AI assistants can produce incorrect information in a confident tone: they may give a source that does not exist, a wrong date or a faulty calculation. Verify the answer against a reliable source, especially for numbers, dates, law, health and finance. Some assistants can search the internet and cite sources; open the source and check that it really contains that information. The assistant's knowledge may only extend up to a certain date, so on current topics it may give incomplete or outdated information.

Privacy: what should you not enter?

  • National ID numbers, health information and personal data belonging to your customers or employees (information covered by data protection law, e.g. the GDPR).
  • Access details such as passwords, bank and card details and API keys.
  • Confidential documents your company has not cleared for sharing: contracts, quotations, financial statements, source code.

Check the data settings of the tool you use (chat history, use of your data to improve the model). Data protection terms may differ on business and enterprise plans; follow your company's rules on the use of AI.

Frequently asked questions

Which AI assistant should I use?

All the widely used assistants can handle basic tasks such as writing text, summarising and answering questions; the differences lie in the interface, the applications they connect to (such as office programs and search) and the terms of use. You can try several with the same task and choose the one that suits you.

Is what I type into an AI assistant stored?

It depends on the tool and your settings. Most tools keep your chat history; some may use the data to improve the model and offer an option to turn this off. Read the terms of use and the privacy settings.

Can I use text written by AI as it is?

Do not use it without checking. Verify the facts and adapt the text to your own style; you are responsible for the text you publish.

Can I write prompts in my own language?

Yes. The widely used assistants understand prompts in many languages and reply in the same language; still, check the spelling and terminology.

BYK Yazılım Support Team
This guide is written and regularly reviewed by the BYK Yazılım support team. Last updated: 4 October 2026.

Related guides

Want to use AI with your own data?

BYK Yazılım builds AI assistants that connect to your business software, so you can ask questions about your production and order data in plain language.

Contact us AI integration