One catalog, every language, one update:
translated, synced, never drifting
A catalog in five languages usually means five separate places where a price, a spec or a label can go stale, because updating one language and remembering to update the other four is a manual step someone eventually forgets. An agent pushes a single update across every language version at once.
The process today
A catalog or a site that exists in more than one language usually started in one language and had translations added later, often by different people at different times, with no shared system keeping them in sync. A price changes, a spec gets corrected, a new product gets added, and updating the primary language is the easy part. Remembering to make the same change in every other language version, and making it consistently, is where the process breaks down, especially once the catalog grows past a size any one person can hold in their head.
A real estate site for a residential development in Montenegro shows what the properly engineered version looks like: a WordPress and Elementor site running in five languages through Polylang, with programmatic content updates pushed through the Elementor API so that every language version changes in a single step instead of five separate manual edits. A supplements factory’s ERP recovery took the same problem into a different domain entirely, translating 3,040 interface strings as part of rebuilding the system self-hosted on PostgreSQL and PostgREST, alongside adding cost accounting and Excel exports.
The cost of staying manual is not just translator hours, it is drift. Teams usually find that the English version of a catalog is the one that actually gets maintained, while the other language versions slowly fall a few updates behind, until a customer in a secondary market sees a price or a spec that no longer matches reality.
What the agent does
The agent takes a source update, a new product, a changed spec, a corrected price, and translates it against a glossary built from your actual brand and product terminology, so a technical term or a brand name does not get translated literally when it should stay fixed. The translation pass is tuned to the tone your catalog already uses, not a generic machine-translation default.
For a CMS-based catalog, the update pushes programmatically into every language version through the CMS’s own API, the same approach used on the five-language Montenegro site through Elementor, so one edit in the source language becomes five synchronized edits instead of five separate manual jobs. For an interface or ERP translation project like the 3,040-string rebuild, the agent works through the full string set in batches, flagging anything ambiguous or context-dependent for a human pass rather than guessing.
Strings that are idiomatic, tied to a regulated claim, or simply ambiguous out of context get flagged separately for a native-speaker review instead of being pushed live automatically, keeping the automation honest about where it is confident and where it is not.
What stays with humans
A native speaker reviews anything flagged as ambiguous or regulated before it goes live, and the glossary of brand and product terms is built and maintained by the team, not inferred by the model. Final sign-off on a catalog or interface going live in a new language stays human.
Guards
A new catalog or string set runs a dry run against existing content before go-live, so the team can check translation quality and glossary adherence before anything touches a live site. Flagged strings never publish without a human pass, every batch is logged against its source update, and a kill switch pauses the sync without losing any already-translated content.
Price and timeline
| Option | Price | What it covers | Timeline |
|---|---|---|---|
| Single automation | from $600 | One catalog or interface into 2-3 languages with a glossary | 5 to 10 days |
| Department package | from $2,500 | Catalog translation plus image alt-text and tagging plus review mining for one product team | 2 to 4 weeks |
Running cost is usually $30 to $90 a month in model usage depending on catalog size and update frequency, with a budget cap set before launch.
Related
Catalog translation pairs naturally with image alt-text and tagging for teams localising a full product catalog, and with review mining for insights once reviews start coming in across multiple markets and languages. See automation everything and AI agents for the broader approach behind both. The five-language programmatic sync is detailed in real estate site in Montenegro, five languages, and the full interface translation project is in factory ERP recovery, self-hosted.
If your catalog’s language versions have started drifting apart, get in touch and we will scope a sync for your setup.
Tired of doing this by hand? We can take the whole routine off your team, not just this step: Routine takeover, from $400 →
FAQ
How much does catalog translation and localisation automation cost?
From $600 for one catalog into 2-3 languages with a fixed glossary, live in 5 to 10 days. Five or more languages, or a full ERP or site interface translation, usually runs $1,500 to $4,000.
How long before it is live?
5 to 10 days once we have the source catalog or interface strings and a glossary of brand terms that should not be translated literally.
Which tools does it connect to?
Polylang and WPML on WordPress sites, a custom CMS through its API, or a database export for an ERP or internal tool. The real estate site in Montenegro runs this through the Elementor API so a programmatic update reaches all five languages in one step.
What happens if a translation is wrong or culturally off?
Strings that are ambiguous, idiomatic or tied to a regulated claim are flagged for a native-speaker review instead of being published automatically, and the glossary locks brand and product terms so they do not drift between languages.
Is our catalog and interface data safe?
Translation runs against your own content and CMS or database, nothing is published without passing through your existing review step if you keep one, and every batch translated is logged against its source.