Top.Mail.Ru
What are prompts and how to write them? — Postmypost
What are prompts and how to write them?

What are prompts and how to write them?

01.04.2024

Read 5 min.
Insights
Ekaterina Novikova

It is difficult to find a field where neural networks have not yet found application. With the help of neural networks, scripts, music, stories, content plans are written, and often truly stunning images are created! However, a quality result can only be achieved if you know how to create prompts for AI.

What is a prompt

A prompt is a set of instructions for a neural network, based on which it generates content. The word comes from English and in the original is written as prompt, which can be translated as "hint". The word has taken root in our language as "prompt" because this pronunciation is easier, but "promt" won't be a mistake either. The quality of the prompt, its wording, and the details of the instructions determine how accurately AI fulfills your request.

How to create a prompt

Understanding how to write prompts for a neural network is not difficult, it's similar to how you enter a query in a search engine: the more specific the information, the better. The neural network is a helpful assistant that requires the most specific instructions for the job. Here are a few recommendations to help achieve the highest quality result in text generation.

  1. First of all, it is recommended to specify the desired content format. What do you want from a neural network: a translation, an event announcement post, a letter? This is what the algorithm will base its response on in the future.
  2. It is also helpful to specify from whose perspective the text is needed, i.e., set a role, as this will also affect the structure and employed words and expressions.
  3. Decide: for what or for whom is this content intended? For example, you need to create an instruction for website users. Based on this, you can also formulate the desired text style.
  4. You can also specify examples and important words that must appear in the text so that the neural network includes them in the final version.
  5. Don't skimp on details and remember the literacy of your request! The more specifics you provide in the prompt, the closer the response will be to the desired result.

Examples of prompts

You can clearly see the difference between quality and poor prompts using examples of those requests we created for the neural network.

Imagine we need to prepare a small message about love for a study at the university. We enter the corresponding request into the neural network and receive a response:

2024-04-01_120528.png

Of course, with the absence of context, the neural network misunderstood the assignment. We will adjust the request, this time adding more details, outlining the text structure, and indicating who is addressing whom:

2024-04-01_120851.png

This result is already much closer to the desired one! However, it still needs your adjustments.

Prompts for content planning and event announcements

The neural network can help with creating a content plan — just provide sufficient data. In the prompt, you should specify for which direction a business plan is needed, for what period, and how frequently the content will be published. You can also add other requirements: columns, from whose perspective to write, and so on. By the way, Postmypost AI focuses on creating not only texts, but also content plans, scripts, and more.

2024-04-01_121527.png

AI can also quickly generate text with event announcements:

2024-04-01_121756.png

The magic formula for prompts

Of course, there is no magic formula, but there are certain rules to follow to write a great prompt and then get a great text from the neural network.

  1. Clearly formulate the topic and purpose of the text.
  2. Select a format for your content: post, article, series of Stories, video script, copy for a landing page, letter, story, meditation, instruction, etc.
  3. Style: specify the style of the text — formal, humorous, restrained, etc.
  4. Volume: indicate the desired length of the future text.
  5. Details: if necessary, specify specific details and aspects that must be highlighted in the text.
  6. Restrictions. Sometimes it's helpful to specify certain limitations. For example, which words or tones to avoid. Or — limit the audience, like: "Write a text for third-year students of the physics and mathematics faculty."
  7. You may need to clarify a number of other details, depending on your tasks.

Prompt builders

If you still have difficulties composing requests, you can use special websites – prompt builders. Such builders offer convenient templates for creating neural network guidelines and entire collections of hints. They also allow you to specify a role and desired format, and the site will offer you to choose the tone and style of the text:

Рисунок5.png

It is important to remember that artificial intelligence is not a living person; sometimes it generates constructions that ordinary people do not use in their speech. Check the resulting text and correct phrases that sound awkward.

You should also double-check the facts provided by the neural network, as it collects information from various sources, not all of which may be verified. Trust but verify!

Learn the art of composing prompts, and the neural network will become a faithful assistant that will help you with tasks at any time, become a source of inspiration and simply relieve unnecessary routine.


Related articles

All posts
How TODO Agency manages its channels through Postmypost

How TODO Agency manages its channels through Postmypost

18.09.2026

Read 4 min.
Insights

Yana, the director of the digital agency TODO Agency, runs channels about her own life, as well as the processes and activities of her company, throug...

Rodionova Julia
Rodionova Julia
Scenarios of crises: from a single angry comment to a viral scandal

Scenarios of crises: from a single angry comment to a viral scandal

17.09.2026

Read 8 min.
Insights

One comment can trigger a chain reaction that cannot be stopped with a simple "Sorry, we'll fix it".  The main mistake is responding to everythin...

Rodionova Julia
Rodionova Julia
Decision point: when it makes sense to abandon one of the platforms

Decision point: when it makes sense to abandon one of the platforms

11.09.2026

Read 5 min.
Insights

In 2026, almost everyone who runs a business or blog adheres to the seemingly perfect approach "You need to be everywhere." Telegram + MAX + VK and mo...

Rodionova Julia
Rodionova Julia
Hypothesis testing in SMM: how to conduct mini-experiments and make quick decisions

Hypothesis testing in SMM: how to conduct mini-experiments and make quick decisions

10.09.2026

Read 6 min.
Insights

Have you ever launched the "perfect" post, but it garnered less reach than a cat meme? The issue isn't with the algorithms—it's the lack of a system....

Rodionova Julia
Rodionova Julia
Checklist: 25 actions for adapting to the 2026 algorithms (VK, Telegram, Instagram*)

Checklist: 25 actions for adapting to the 2026 algorithms (VK, Telegram, Instagram*)

01.09.2026

Read 8 min.
Insights

If you have read our article about 8 main changes in the algorithms of VK, Telegram, and Instagram*, you already know: likes no longer work. Now algor...

Rodionova Julia
Rodionova Julia
Channel content audit: how once a quarter to find "sleeping" materials with potential

Channel content audit: how once a quarter to find "sleeping" materials with potential

31.08.2026

Read 4 min.
Insights

Most channels live by the principle "published - forgotten". Meanwhile, the archive almost always contains posts that still bring transitions from sea...

Rodionova Julia
Rodionova Julia
All posts