Skip to content
FAS Localize
All topics

AI localization pipelines

An AI localization pipeline watches your content, drafts translations with several models, lets a judge model and quality checks pick the best, routes only uncertain segments to people, and ships the result back automatically.

These articles explain how that works in plain terms, compare models and pipelines on the market, and track what changes in the field each month.

All AI pipeline articles

Comparison · Nov 20, 2026 · 5 min read

Do TMS Platforms Support AI Agent Automation?

Do TMS platforms support AI agent automation? Crowdin, Phrase and Lokalise each shipped a trigger that lets an agent or webhook start a translation workflow.

News & views · Nov 16, 2026 · 5 min read

How Many Steps Does AI Translation Need?

Intento's 2026 benchmark of 229 AI translation solutions in 100 languages shows how many pipeline steps each quality bar needs, and what every extra step costs.

Guide · Nov 12, 2026 · 4 min read

What Is FACET Translation Quality Evaluation?

FACET is a 2026 method for translation quality evaluation: it checks fluency, accuracy and consistency in three passes, each given only the context it needs.

Before / after · Oct 23, 2026 · 4 min read

One Model for QE and Post-Editing

A unified 0.5B model scores and fixes translations in one pass, but a translator study found smarter highlights alone don't speed up post-editing.

News & views · Oct 19, 2026 · 5 min read

Error-Free Segment Detection in Translation

WMT26 made error-free segment detection its own task. See how pipelines now decide a translated segment needs zero human review, with no reference text.

Comparison · Oct 15, 2026 · 5 min read

LLM Judge vs COMET for Translation: What WMT25 Found

LLM judge vs COMET for translation: WMT25 found LLMs rank MT engines well but lose to reference-based metrics on single sentences. How to use each one.

How-to · Oct 11, 2026 · 4 min read

How to Translate a Shopify Store with AI: The API Route

How to translate a Shopify store with AI through the Admin API: enable a locale, read digests, draft, score with QE, write back and publish the language.

Comparison · Oct 7, 2026 · 4 min read

Reference-free vs reference-based QE: what each measures

Reference-based metrics score a translation against a human reference, while QE without references judges it alone. Here is when each one fits best.