Using modern AI models — such as GPT, DeepSeek, Mistral — took website translation automation to a completely new level. This is especially relevant for Tilda websites, where multilingual support is not available out of the box, and automatic translations from Google, Yandex, etc. require manual adjustments, which is a labor-intensive and error-prone process, and also doesn't provide the desired results for SEO optimization, as the site is translated after it's loaded in the browser.
The Multify service Multify translates text not just with single requests to AI models but uses a whole system — taking into account context and graph architecture, which helps maintain text coherence across the entire site. Here's how it all works.
🔄 Why context is important
When translating text on a website, especially a small fragment (e.g., a menu item, button, or footer line), an isolated approach yields inaccurate results. The model may not understand what the phrase refers to, how it agrees with other elements, and may choose the wrong translation.
To avoid this, Multify provides the model not only with the text to be translated but also its context:
[ preceding text ]
[ text to be translated ]
[ following text ]This approach helps the model to “see” the fragment not as a disconnected snippet but as part of a coherent whole.
📍Example
On this site, when translating the button "more about" the model takes into account the context of the upper and lower blocks:
In the code below, the considered context around the button is highlighted "more about":
➰ Graph Structure: How Context is Formed
To more accurately determine the environment of fragments, Multify breaks down the entire document (web page) into blocks and forms a bidirectional graph from them. This means that:
- Each text element knows which blocks are nearby.
- If a new fragment is added in the middle of the page, its "neighborhood" can be automatically determined.
- The model receives not only the fragment itself but also logically related blocks — even if they were translated earlier.
This approach helps maintain semantic integrity as well as grammatical consistency — for example, correct cases, tenses, and stylistic uniformity.
💯 Why does it work better?
Context in translation is the key to quality. This is especially noticeable in:
- Complex names and technical terms
- Short phrases without verbs (e.g., «For Home», «To Warehouse»)
- Repeating elements that depend on the environment
Without understanding the context, LLM can generate a “formally correct” but unnatural or incorrect translation. Thanks to the graph and correct transmission of the environment, Multify avoids these errors.
🔝 What is the difference for SEO?
In addition to the quality and accuracy of translation using AI models, there is also a significant difference in the technical implementation of a multilingual website, which greatly affects search engine rankings. The thing is, translating a website using Google Translate or Yandex Translate, which is done after the page is loaded in the browser, such translations are performed on the client-side and do not contribute to SEO optimization.
Search engines do not index translated content, created using automatic tools without human editing, as it is considered automatically generated content.
Search engines do not index translated content, created using automatic tools without human editing, as it is considered automatically generated content.
"Google does not index translated content created using Google Translate, which limits the visibility of your website in international markets."
→ Source: Auris AI
[my translation]
Thus, for effective SEO-optimization of a multilingual site it is recommended to use server-side solutions, which provide search engines with access to translated content.
🚀 Claude — best translation quality
Currently, Multify we are using Claude3.5 Haiku — this LLM shows the best translation quality for languages of CIS countries: Kazakh, Uzbek, Ukrainian, Romanian, Azerbaijani, Kyrgyz, Armenian, Tajik, Belarusian, Turkmen.
Despite the high cost of Claude compared to other AI models, thanks to our proprietary architecture and the use of unlimited servers, Multify offers:
💸 Competitive pricing
🥇 The highest quality among solutions for Tilda
🌐 Support for complex languages and regional variations
🎯 Summary: how it all works together
- The page is broken down into logical chunks
- Each chunk gets its context — the text before and after
- A graph is formed, so that when updates are made, the surrounding context can be quickly identified
- The model gets the necessary context and produces a coherent, accurate translation
As a result, you get:
✅ Multilingual Website without duplicates
✅ Translation that reads naturally
✅ Improved SEO tags and meta tags
✅ Flexibility and scalability without manual routine
✅ Translation that reads naturally
✅ Improved SEO tags and meta tags
✅ Flexibility and scalability without manual routine
