Home › Artificial Intelligence for All › Effective Interaction

Effective Interaction

Learn how to write effective prompts to maximize the capabilities of Large Language Models.

The ability to write effective prompts is a critical skill in the modern digital era, as it determines the quality of interaction with LLMs. A well-formulated prompt leads to more accurate, relevant, and useful outputs, whereas a vague or incomplete prompt can result in generic or irrelevant responses. Cultivating this skill is the primary objective of this guide. Drafting effective prompts involves providing sufficient information, defining clear objectives, and specifying expectations regarding the content, tone, and format of the response.

Clarity

Clearly state what you want to achieve

Structure

Organize the prompt into logical sections

Context

Provide sufficient context and examples

The following sections describe basic principles and advanced methods for writing prompts.

Why is it important?

Developing prompt writing skills contributes to saving time and resources. Instead of needing multiple attempts to achieve the desired output, a well-designed initial prompt can lead directly to useful and immediately applicable content. This skill becomes particularly important as LLMs are increasingly integrated into professional, educational, and everyday tools.

Basic principles of writing prompts

An effective prompt often consists of distinct sections that define the parameters of the task. It is not necessary to use all sections in every request, but combining them significantly improves the accuracy of the response. The basic elements include defining the role, providing instructions, describing the target audience, and specifying the response format. Additionally, a prompt can include attached files (e.g., texts, images) or links that provide supplementary data for processing.

To facilitate the user, an indicative prompt template is proposed below. It is emphasized that this is just one of many possible structures, as the method of effectively communicating with these systems is a constantly evolving research field. Below, we use the “#” symbol to denote headings, to distinguish and make the information we want to include in the prompt as clear as possible. This convention is based on the Markdown text formatting language.

Basic prompt template

# ROLEYou are a [role] [with expertise in a field and specific skills].
# INSTRUCTIONS[Active verb] [specific task] [desired output].
# AUDIENCE[target audience / tone]
# FORMAT– Length: [number of words/pages]– Formatting: [e.g., bulleted list, 3-column table]
ATTACHMENTS: documents, spreadsheets, images, etc.
1

Role

Define the expertise

Assigning a role to an LLM (e.g., “business promotion expert”) helps it adapt not only the content but also the manner of its response. Specifying a role activates the appropriate tone and level of expertise.

Note: The “role” acts as a technical parameter that directs the LLM to focus on specific functions. This is an advanced simulation of patterns, not an expression of personality.
2

Instructions

Define the task

Instructions form the core of the prompt and define the task to be executed. They must be formulated clearly and accurately.

Good practice“List 5 practical ways to reduce plastic waste in the office “
Avoid“Write about recycling”
3

Audience

Define the recipients

Defining the target audience (e.g., “students”, “business executives”, “general public”) determines the level of complexity, terminology, and tone to be used in the response. A text aimed at experts may include technical terms, whereas a text for non-experts requires simpler language.

4

Format

Define the output structure

Defining the output format sets the structure of the response. Controlling the formatting makes the output more readable and ready to use.

Bulleted list 100-word summaryTwo-column tableThree paragraphs

Attachments

Add context and data

Attaching files (such as text documents, spreadsheets, or images) or pasting data directly into the prompt allows the LLM to process information provided by the user.

Tip: For better organization, leave blank lines and use a separator symbol like “—” between the instructions and the data.

Advanced prompting methods

For more complex tasks, the basic structure of the prompt can be enriched with additional elements that increase the accuracy and control of the generated output. These elements include defining criteria, breaking down the task into sequential steps, and providing examples.

Advanced prompt template

# ROLEYou are a [role] [with expertise in a field and specific skills].
# INSTRUCTIONS[Active verb] [specific task] [desired output].
# AUDIENCE[target audience / tone]
# FORMAT– Length: [number of words/pages]– Formatting: [e.g., bulleted list, 3-column table]
# CRITERIA– [criterion 1]– [criterion 2]
# STEPS1. [first step with a specific instruction]2. [second step with a specific instruction]
# EXAMPLES## EXAMPLE 1Input: [Sample input]Response: [Sample desired response]
5

Criteria

Set constraints and quality rules

Examples of criteria:

  • Word limit
  • Mandatory use of examples

Additionally:

  • Avoidance of specialized terminology
  • Neutral and objective tone
6

Steps

Break down into distinct actions

For complex tasks, breaking down the prompt into distinct, numbered steps directs the system to follow a specific logical sequence. Segmenting the task into steps reduces the probability of errors or omissions.

Example:

  1. Identify the main arguments in the text
  2. Write a 50-word summary
  3. Translate the summary into English
7

Examples

Demonstrate the desired pattern

Providing examples of desired inputs-outputs within the prompt is an extremely effective technique for guiding the system. It acts as a template indicating the exact pattern, tone, structure, or format expected.

Categorization example:

INPUT“Excellent service!”
RESPONSEPositive
Practical Examples