This episode of the Jukebox podcast features Leonardo (Leo) Losoviz, a long-time WordPress developer and author of Gato GraphQL and Gato AI Translations for Polylang. He explains how recent AI advances make translating WordPress sites faster, cheaper, and easier — and what site owners and developers should watch out for when adopting those tools.
Why translate your site?
Leo separates the cases: if local law requires multilingual content, you must comply. Outside of legal obligations, translating is simply good business when you can reach users who speak other languages. With AI making translations cheap and fast, there’s little reason not to offer multiple languages when it makes sense for your audience.
How AI changes the translation equation
AI translation quality has improved dramatically. For general content — blogs, marketing copy, product pages — modern models often produce excellent results. The main remaining gaps are domain-specific or technical terms, acronyms, and ambiguous phrases where context matters.
Cost is a major shift. Instead of paying professional translators hours of work per page, many sites can now auto-translate for the price of a few cents in API tokens. That doesn’t eliminate the need for human review in some cases, but it reduces the time and expense of getting to a near-finished translation. Leo recommends using professionals selectively to validate industry-specific language or important legal text.
When a human still matters
If your site uses specialized terminology (legal, medical, scientific) or must be 100% precise for compliance or brand reasons, engage a human translator to review or polish the AI output. With AI you can reduce a translator’s role to verification and fine-tuning rather than full draft production, cutting cost and turnaround time.
How Gato AI Translations works (overview)
Leo’s plugin, Gato AI Translations for Polylang, hooks into Polylang’s approach: when you translate a post it creates separate post entries for each target language. From the post list you can bulk-translate posts into one or many languages; the plugin duplicates the post(s) and inserts AI-generated translations for title, content, tags, categories and image metadata.
This static-per-language approach has advantages: pages can be cached and served quickly, and you avoid expensive runtime string lookups. Polylang also supports language-specific URL structures (example.com/fr/post-slug or fr.example.com), and adds hreflang markup so search engines know which pages are translations of each other.
Alternatives use different strategies — WPML, TranslatePress, Weglot and others vary between runtime translation and static duplicates. Choose what fits your site’s performance and workflow needs.
UI and workflow
Leo favors minimizing the need to work in multiple language UIs. With his setup you primarily edit the origin post (your master language) and trigger translations from the post list. If you must review a translated version, you open that specific post and adjust it in the WordPress editor.
WordPress 7.0, AI connectors and collaboration
WordPress is adding infrastructure that will make AI and collaborative workflows easier. The upcoming AI Connector will standardize connecting editor tools to AI services; collaborative editing (phase three of Gutenberg) introduces comment-like interactions inside the editor. Together these will enable scenarios where AI can flag uncertain translations and create in-editor notes or suggestions that editors can review, similar to Google Docs comments but involving AI as a participant.
Best practices and a simple checklist
– Finalize content before translating: don’t translate drafts that will be changed repeatedly. Each extra round wastes tokens and review time.
– Check structure and assets: confirm headings, alt text, captions, and image placement are correct. Avoid embedded text inside images where possible.
– Review media and embeds: embedded foreign-language videos or infographics may not make sense after translation and may need replacement or context notes.
– Use AI for volume, humans for precision: let AI do most of the heavy lifting; have humans fix edge cases.
– Translate metadata: captions, image titles and alt text should be translated where appropriate — Polylang creates language entries for media metadata without duplicating the actual image file.
Images, captions and non-text content
Technically it’s feasible to localize images (replace or recreate infographics with localized labels), but Leo warns against blindly duplicating every image for every language — that quickly balloons site complexity. A leaner approach: keep images language-neutral when possible (no embedded text), and place translatable text overlays or captions that are handled like normal post text. Polylang allows translated metadata for media entries without duplicating the binary file.
SEO and discoverability
When done correctly, multilingual sites improve discoverability in target languages. Polylang and similar plugins provide hreflang tags and distinct URLs so search engines treat language versions as related rather than duplicate content. However, because AI lowers the barrier, many sites will go multilingual; this levels the playing field and can become an arms race — you may need translations simply to remain competitive, not just to gain advantage.
Operational advice
– Keep the origin post language as your working view; translate only after the origin is finalized.
– Use bulk translate actions for efficiency, then selectively review the most important pages.
– Track where human review is necessary (legal, technical, brand-critical pages).
– Monitor how competitors adopt translations and adjust priorities accordingly.
The broader picture
AI makes multilingual publishing accessible for almost any WordPress site. The remaining challenges are process and oversight: knowing when to rely on AI, when to involve people, and how to structure content and assets so translations are maintainable. WordPress’s evolving editor and AI integration will make collaboration and review more natural inside the CMS, but community contributions and plugins will flesh out the real-world use cases.
If you want to explore this further, check Leonardo Losoviz’s plugin (Gato AI Translations for Polylang), watch his WordCamp Asia talk for deeper details, and follow WordPress core updates around the AI Connector and collaborative editor features. The practical takeaway: if expanding to other languages helps your audience or keeps you competitive, AI has made doing so far easier and far cheaper — but plan your workflow to avoid wasted effort and ensure quality where it matters.