AI and content terms explained: prompts, parsing, rewrite and tokens
Plain-language explanations of AI and content terms with Kotia examples: rewrite vs parsing, context vs memory, model tokens vs your balance, and workflows.
On this page5
This section helps you understand which tool you need for your task and what its settings mean. For example, page parsing gets text from a link, rewrite reworks the wording, and fact-checking checks claims. These actions can work together, but they produce different results.
Each article in this section has a short explanation, an example, and the steps to follow in Kotia. Choose the question you have right now.
How to give the AI a task and work with large materials
| Question | What to read |
|---|---|
| How do I tell the AI what result I need? | Prompt — the task with source data and requirements |
| How does the assistant search and perform other actions? | AI agent — a model that uses the tools available to it |
| Why can't I add as much text as I want to one conversation? | Context window — the amount of data the model can use when it answers |
| How is text length related to usage? | Tokens — units of text processing and how they differ from your Kotia balance |
| How do I repeat several steps without programming? | No-code workflow — building a chain from cards |
Where to find topics and source material
- Keywords help you choose topics based on search queries.
- Content plan connects topics, formats, and the publishing order. It is a plan for several materials, not the structure of a single article.
- Page parsing gets the accessible text of web pages for further processing.
- Deep Research studies a topic using sources and prepares a report with conclusions and links.
- Fact-checking helps you check claims in a finished material.
How to prepare text for your task
| Action | What it gives you |
|---|---|
| Copywriting | Text for a specific audience and business goal |
| Rewrite | A new version of existing material that keeps the meaning |
| Tone of voice | A communication style that makes your posts recognizable |
| Digest | A summary of several materials that highlights the main points |
Example: prepare a post based on a blog article
You found a topic in your favorite industry blog that you want to cover for your audience. In Kotia, this process can be broken down into a few clear steps:
- Pass the link and get the text through page parsing. At this stage, you get the source material, not a new post.
- Write a prompt: who the post is for, which idea to develop, which facts to keep, and how to cite the source.
- Ask the AI to prepare the text. If you need to rework the original wording, that is a rewrite. If you need your own analysis, add your opinion and requirements for the new material.
- Apply the style you need — tone of voice — and read the result. Check the important claims and links.
- If you work this way regularly, build a workflow to pass results between steps without copying them by hand.
You can set up this workflow step by step in the article on Telegram posts from a link. To choose a different task, open the catalog of ready-made workflows.
Terms that are easy to mix up
Context and memory. Context is the data the model can use for a specific answer. Memory saves individual details for later work. It does not mean that any text from an old conversation is always in front of the model. For more, see how memory works.
AI detector and fact-checking. The detector assesses signs of where a text came from. Fact-checking checks claims against sources. A confident score from the detector says nothing about whether the figures in the article are accurate.
Workflows and publishing. A chain can only prepare and save material. To send it to a website or a channel, you need a separate action built for that purpose and its setup.