AI translation and localization services
We run AI translation and localization as a managed service, inside your files or your platform. You can use it in two ways: we translate your new content, or we assess AI output you already have.
Our five services share your language resources and our quality controls, with a workflow matched to each service. Your quote lists the steps and deliverables included. One team stays on your account, so your glossary, translation memory (TM), style rules, and reviewer feedback carry across everything we do for you, from help articles to product UI, and a native linguist works on every job.
Every delivery includes the quality-gate result and an issue log. At AI + Edit, sampled quality reports follow the schedule agreed for your program and can cover several small releases; at AI + Edit + Review, each delivery gets a full report.
High-volume programs are normal business for us. Send a single batch or a continuous stream of updates.
AI translation with linguist editing
AI translates your files, and a native linguist works on them at one of three depths: a linguist check at AI + Check, full post-editing at AI + Edit, or full post-editing plus a second linguist’s review at AI + Edit + Review. At AI + Edit and above, this is AI post-editing (AIPE), and it adds more than a typical machine translation post-editing (MTPE) workflow: the AI is calibrated on your rules before it translates, agents check the text before and after the post-editor, and a quality gate checks every delivered file. You get the edited file and the evidence of what was checked.
| You send | You get back |
|---|---|
| Files in your format (for example, DOCX, XLSX, CSV, XML, JSON, XLIFF, HTML, or PO), or a job in your translation management system (TMS), plus your glossary, TM, and style guide if you have them | Files in your format, the quality-gate result, and an issue log with every delivery, plus, at AI + Edit and above, glossary changes, a query log, and quality reports on your level’s schedule |
Continuous localization for websites, apps, and catalogs
Every language stays in step with your release cadence, and text that has not changed is not translated again. We compare each source release with the previous source version. We translate what changed and recheck existing translations affected by changes in context, approved terminology, or product behavior. We ask for context such as screenshots, string keys, and length limits, and we work inside your platform, including Phrase, XTM, Smartling, Crowdin, and Lokalise. Product catalogs and e-commerce content run here too. On request, we localize catalog titles for the search terms shoppers use in each market.
| You send | You get back |
|---|---|
| Each release in your TMS, or as an export | The changed content in your format with each release, the gate result, and an issue log. At AI + Edit and above, also glossary changes, a query log, and quality reports on your level’s schedule |
Help-center and support content
Your knowledge-base articles, support macros, and other help-center content stay in step with your source articles. We translate what changed, recheck existing articles the change affects, keep product terms consistent with your UI, and set the depth by article priority. Edits your team makes after publication feed back into your glossary and rules.
| You send | You get back |
|---|---|
| Article exports, or a job in your TMS, on your update cycle | Updated articles in your format, the gate result, and an issue log. At AI + Edit and above, also glossary changes, a query log, and quality reports on your level’s schedule |
Quality review of your AI or MT output
You already have AI or machine translation (MT) output and want to know whether it holds up before more of it goes live. A native linguist, supported by quality assurance (QA) agents, assesses it and classifies every issue as critical, major, or minor, by category. The report recommends a service level for the content. Correction is quoted separately.
| You send | You get back |
|---|---|
| Source text and the AI or MT output, aligned, plus your glossary or style guide if you have one | A report with issue counts, an issue log, the main patterns, and a recommended level |
Glossary, style guide, and translation memory
Consistent AI translation starts with clear rules. If you have no glossary or style guide yet, we create them with you before production starts. Every term carries its source, a native linguist approves it, and you approve the result. We can also align your past translations into a TM. Our agents translate against them, and the rules that can be tested reliably become automatic checks in the quality gate.
| You send | You get back |
|---|---|
| Your published content, past translations, and brand guidelines | A sourced glossary in your format, a style guide, and a Translation Memory eXchange (TMX) file |
Choosing between our localization services
Start with what you already have.
- You have AI output and doubts about it. Start a free check. We review up to 300 source words of your existing AI or MT output in a language pair we offer, and send you the findings.
- You have content waiting to be translated. Request a quote and list your content types. The quote names a service level for each.
There are two ways in. If you already have AI or MT output, start with a free check. If you have content to translate, request a quote, and if the fit is right, we run a sample on your own file. A sample tests a defined part of your content. A pilot tests delivery under real working conditions. Before either starts, we agree the scope, deliverables, timing, and any charge. Your quote confirms whether the work is a one-off project or part of an ongoing program.
Human-led projects and DTP at locstars.com
This site covers AI translation and localization for high-volume content. If you want a project led by human translators from the start, or you need desktop publishing (DTP), visit locstars.com, the main Locstars site. On this site, you can also choose no AI for any content type in your program. That content is translated by native linguists without machine translation or generative AI, using your translation memory and glossary.