The release of version 2 of the Polylang AI Automatic Translation plugin marks a big leap forward for multilingual website management. After years of focused development, the new version brings in features that improve translation accuracy and give users more control.
For developers using Bricks Builder, this update means deeper integration and smoother workflows. Managing content across multiple languages just got a whole lot easier—and more consistent—for both site creators and visitors.
Why We Rebuilt Everything
They started out building separate integrations for each page builder—Elementor, Divi, and Beaver Builder—and each one needed its own codebase. This worked for a while, but the maintenance was never-ending.
Results from version 1:
| Builder | Success Rate | Issues Found |
|---|---|---|
| Elementor | ~90% | Edge cases, custom widgets |
| Divi | Partial | Layout inconsistencies |
| Beaver Builder | Partial | Unsupported modules |
That missing 10% pointed to a deeper problem. Every builder brought new exceptions, custom elements, and third-party add-ons that needed individual fixes.
It became clear that writing code for every weird scenario just wasn’t going to scale.
To move forward, they needed a unified system—one that could handle all these variations automatically instead of patching them one by one.
The V2 Breakthrough: 100% LLM Intelligence
Version 2 completely changed the approach, relying on the language model’s reasoning instead of hard-coded instructions. No more assigning fixed rules to each builder—the LLM now interprets and manages structural data by itself.
Modern models like ChatGPT, Claude, and Gemini use specialized tools that expand what they can do. These tools fill two main roles:
| Purpose | Function |
|---|---|
| Builder Structure Handling | Guides the LLM to identify, read, and save content for different page builders, even with unique JSON layouts. |
| Translation Retrieval | Lets the LLM fetch translations, compare updates, and translate only what’s changed. |
This setup removes repetitive work and leads to more precise updates. The LLM doesn’t just follow recipes—it interprets context, sees how elements relate, and decides what needs to change. You get a workflow powered by actual understanding, not just mechanical rule-following.
Why This Reaches 99% Instead of 90%
Version 2 ditches fixed code logic and lets the language model read layouts directly. No more depending on rules for every widget or plugin. The model scans the structure and figures out how each element fits.
Key improvements include:
| Area | v1 Limitation | v2 Enhancement |
|---|---|---|
| Structure Handling | Only standard layouts worked | Recognizes flexible and custom structures |
| Context Awareness | Translated elements separately | Understands full-page context |
| Instruction Control | Limited customization | Applies user-specific translation rules |
Now, the system adapts to unexpected or mixed setups. It recognizes third-party components, nested designs, and user preferences—no manual code updates needed.
By leaning on the model’s contextual reasoning instead of rigid templates, the process covers almost every scenario.
What This Means for Bricks Users
Bricks users now get a smoother translation process that fits their existing layouts—no extra setup required. The system spots Bricks pages automatically and reads the internal structure before translating. No need to map things out by hand.
Key improvements include:
| Feature | Benefit |
|---|---|
| Automatic Recognition | Detects Bricks content and applies accurate translation rules. |
| Layout Awareness | Handles nested containers, loop grids, and custom elements correctly. |
| Dynamic Data Support | Keeps queries, ACF fields, and meta values working after translation. |
| Design Integrity | Maintains spacing, colors, and typography while only translating text. |
Users keep working with their custom designs and data like always. The translation process respects each page’s structure and function, saving time and reducing layout or content mistakes.
Smart Re-translation That Saves Time
When you update content—edit a headline, add a section—the system finds exactly what’s changed. It doesn’t reprocess the whole page; it just zeroes in on the new or edited parts, leaving everything else alone.
Process Overview:
| Step | Action | Result |
|---|---|---|
| 1 | Retrieve the previous translated version | Maintain reference consistency |
| 2 | Detect text differences | Target only updated content |
| 3 | Translate new or edited parts | Save time and API usage |
This targeted method keeps translations accurate and avoids unnecessary rework.
One Framework, Every Builder
The updated framework translates content across every WordPress builder in a single, consistent system. It figures out whether your content comes from Bricks, Gutenberg, Elementor, ACF, or WooCommerce and applies the right logic on its own.
No more switching between tools. The AI reads the structure, grabs the right instructions, and completes translations—no manual setup needed.
| Builder Type | Example Use |
|---|---|
| Bricks | Main site pages |
| Gutenberg | Blog posts |
| Elementor | Legacy layouts |
| ACF | Custom post types |
| WooCommerce | Product pages |
Intelligent detection, not fixed code, makes the framework adapt to complex or mixed setups. Builders work together, and translations stay consistent throughout the site.
Bulk Translation Built for Real Sites
Big sites need translation tools that can handle scale without falling apart. Bulk translation kicks in with one click, and the background processor takes care of every page automatically.
| Feature | Benefit |
|---|---|
| Real-time dashboard | Tracks completed and pending translations |
| Automated scheduling | Prevents timeouts and crashes |
| Smart re-translation | Updates only modified content |
Teams can update dozens of pages at once, keeping every version aligned—no extra manual work.
Why Development Spanned 2.5 Years
The team didn’t want quick fixes—they set out to build a complete, reliable system instead of a half-baked solution. Early tests showed that fast, custom code could get about 90% accuracy, but that still left too many weird, unpredictable errors.
They pushed for a framework that could handle custom setups, third-party tools, and different user workflows without breaking. That meant designing a structure where the AI could actually interpret data, not just follow instructions.
| Goal | Outcome |
|---|---|
| Short-term release | 90% reliability |
| Full AI-driven framework | ~99% reliability |
That shift—from rule-based code to genuine AI reasoning—explains why it took 2.5 years to get here.
Ready for Your Bricks Sites
Version 2 now supports Bricks Builder for multilingual setups using Polylang. It handles complex layouts and custom elements with accurate translation results.
You can combine Bricks with other tools and still keep everything compatible. Even third-party extensions play nicely.
Key Highlights:
- Works seamlessly with unique site configurations
- Maintains translation accuracy across custom designs
- Integrates with other builders and plugins
| Feature | Benefit |
|---|---|
| AI Translation v2 | Understands WordPress structure |
| Bricks Compatibility | Preserves layout integrity |
| Multi-Tool Support | Expands workflow flexibility |
Now, developers can apply automatic AI translation straight to Bricks sites. Go ahead—test performance in real projects and see how it holds up.