Frequently Asked Questions

kotia.app is a platform for creating quality content using artificial intelligence. We offer three main tools: long-form article generation with research and fact-checking, deep research on any topic analyzing dozens of sources, and news creation adapted to your style and audience. The project is built with care — every feature is crafted for quality.

You can create three types of content. Articles — SEO-optimized long-form pieces (500–2000+ words) with source links and fact-checking. Research — deep analysis of any topic using Gemini Deep Research, analyzing dozens of sources over 5–60 minutes. News — find current news and create content in your style for your blog or publication.

Yes, all articles are always unique and generated from scratch. We don't copy or paraphrase existing content — each article is created based on original research through multiple stages of writing, editing, and fact-checking.

Most likely — no. The text is validated by different AI agents at different stages. The probability that one agent writes something questionable and several others don't catch it is extremely low. Our multi-stage pipeline ensures natural, high-quality output.

The interface is fully localized in 6 languages: English, Russian, German, French, Spanish, and Portuguese. You can create content in any of these languages. Research results and AI thinking steps are also translated in real-time.

Use the built-in Support Chat — click the chat icon in the interface. You can also email us at hello@kotia.app or write to our Telegram. We read and respond to every message.

There's one assistant that you talk to in plain language. The home screen offers four services — Research, Article, News, and Agent — each with a library of ready-made prompt templates to help you get started. The assistant decides on its own which tools to use based on what you ask, so you don't need to choose a mode upfront.

Article generation takes 15–25 minutes — 6 sequential stages: planning, source search, evidence gathering, structuring, writing, and final editing. Each stage takes 2–5 minutes. This approach ensures high-quality, well-researched content.

Yes! You can refine any article through the chat. Simply describe what you'd like to change — add a section, change the tone, update information, expand a topic. The system saves up to 5 versions of each article, so you can always go back to a previous version using the version dropdown in the article view.

Deep Research uses Gemini's advanced capabilities to thoroughly analyze a topic. It searches and analyzes dozens of sources over 5–60 minutes, showing you its thinking process in real-time. You can also request follow-up research to expand on specific aspects. Just describe your topic in chat — the assistant will run the research.

Just describe what news you're looking for in chat. The assistant searches current news via Google RSS, shows you the sources found, and creates the selected news in your style. You can customize the author's tone, target audience, and publication name.

Yes! Batch generation lets you create 2 to 20 articles in one go. Set up your topics and parameters, and the system will generate all articles simultaneously. You can track the progress of each article individually. Tokens are charged per article based on the options chosen.

Yes. Every time you refine an article, the previous version is automatically saved. Up to 5 versions are stored (oldest are replaced when the limit is reached). You can switch between versions using the dropdown in the article view.

On any message in the chat, hover over the three-dot menu and select "Report". Describe what went wrong — we review every report personally. If the system produced clearly unsatisfactory results due to an error, we will refund the tokens spent on that operation. You can also use the Support Chat to discuss quality concerns directly with our team.

A board where you place the steps of a job and wire them into a route: search sources, write, fact-check, save to a document. You build it once and run it as often as you like. No coding at any step. The section opens on a computer, from 1024 pixels of window width.

There is no fixed price: a run costs the sum of what its cards spend. Cards that only move data around are free; a card that asks a model, or launches one of our services, costs what that work costs — for a service card, the service plus the agent turn that fills it in. The board does not show a total in advance, so start with a short scheme if you want to see the scale.

Building a scheme means dragging cards and drawing wires between them on a large surface. A phone screen has no room for that, so on a narrow screen the section is hidden and its address explains that it needs a computer.

An estimated cost is charged when you confirm an operation. After completion, the difference is refunded if the actual cost was lower. Approximate costs: chat message — ~1 token, chat with Web Search — ~12 tokens, article generation — ~90-250 tokens (estimate: 150), Deep Research — 250-400 tokens (estimate: 250, varies by complexity), news search — 5 tokens, news creation — 10 tokens, article refinement — ~15 tokens. The AI check is the exception to the estimate-and-refund rule above: it is priced by text length, the price is known before the check starts and is never recalculated — about 5 tokens for 500 characters and about 450 for the maximum of 50,000.

Each day we top up your bonus token balance to 100: if you have fewer than 100 bonus tokens, the difference is credited at midnight UTC. If you already have 100 or more, no top-up happens. Bonus tokens are spent last — first the subscription quota, then purchased tokens, then bonus ones.

Purchased and bonus tokens never expire. Subscription quota tokens do: they are valid until the end of the paid period and do not carry over to the next one. Spending order is fixed — subscription quota first, then purchased tokens, then bonus ones.

Card payment via YooKassa (in rubles) is available to visitors from Russia — both for token packages and for the subscription. Everyone else pays with cryptocurrency (USDT TRC20) right on the checkout page, no need to write to us. Stripe is not connected.

Larger token packages include generous bonuses: Mini (1,000 tokens, no bonus), Standard (4,000 + 400 bonus tokens, 10% extra), Optimal (10,000 + 1,500 bonus tokens, 15% extra), Maximum (20,000 + 5,000 bonus tokens, 25% extra). The bigger the package, the more bonus tokens you receive — up to 25% extra.

Every new user can test the service using bonus tokens (100) completely free. If you purchased tokens and want a refund, it's available within 14 days of purchase for unused tokens. Bonus tokens are non-refundable. Request a refund through the Support Chat or by contacting us directly.

Yes. Choose cryptocurrency on the checkout page — the payment page with the amount and the wallet address opens automatically, and tokens are credited once the network confirms the transfer.

The price of a check depends on the length of the text: about 5 tokens for 500 characters, about 23 for 2,500, about 45 for 5,000 and about 450 for the maximum of 50,000. The number is shown next to the button before you start, and exactly that number is charged — nothing is recalculated afterwards. One check takes from 500 to 50,000 characters. The check itself is available with an active subscription; if a check fails, the tokens are refunded in full.

The platform uses several AI models: Gemini, Claude, Codestral, and DeepSeek. The system automatically selects the optimal model for each task and has fallback to alternative models for reliability.

Yes. Web search is built into the assistant and turns on automatically whenever your question needs fresh data — current news, statistics, recent events, or anything time-sensitive. You don't need to enable anything yourself. A web search costs approximately 12 tokens per query.

Yes, your data is stored securely in an encrypted database. We don't share your data with third parties under any circumstances. Your generated content belongs entirely to you. We use secure HTTPS connections and follow data protection best practices.

An account stores up to 50 MB of files without a subscription and up to 1 GB with one — this covers both chat attachments and the images the chat draws. Nothing is ever deleted automatically: when the space runs out, the upload is refused with a clear message, and you free space yourself or get more. If you delete a file and the assistant needs its content again, that turn simply continues without it.

You can register with Google (one click) or with email and password. Email registration requires email verification. Either way, you'll immediately receive 200 welcome tokens to start exploring the platform.

All your created content is available on the History page. You can filter by content type (articles, research, news), status, and language. Click on any item to view the full result, switch between article versions, or continue working in the chat.