Translation Engine & Translation Memory

Overview

ToRun includes a dedicated translation engine covering roughly 100 languages, plus a translation memory that stores prior translations so repeated phrases stay consistent and don't have to be re-translated from scratch.

Translation runs as its own job type, so you can translate a phrase, a document, or a stream of text and track each job's status.

Translating text

1. Open the translator surface (or use translation as a workflow step / tool).
2. Pick the source language (or let it auto-detect) and the target language.
3. Submit. A translator job is created and processed; the result is returned with the translation.

Because translation is an AI capability, each translation call is billed like any other call — one BillingRecord per call with a pricing snapshot.

Translation memory

Translation memory (TM) keeps a record of segments you've already translated. When the same or a very similar segment comes up again, the stored translation is reused. This gives you:

• Consistency — the same term is translated the same way every time.
• Lower cost and latency — reused segments don't need a fresh model call.
• Quality auditing — translations can be reviewed and scored over time.

TM is organized around translation groups and resources, and quality audits can be attached to translations to track review status.

Languages and localization

• The translation engine targets ~100 languages.
• The ToRun product UI itself ships in 29 supported languages; translation lets you work across far more than the UI locales.
• Country, currency, and timezone catalogs back locale-aware formatting.

Cross-language chat (mobile)

A related feature lets two people chat as friends with each side writing and reading in their own language, with two-way automatic translation between them. This is part of the mobile experience and builds on the same translation engine.

Practical tips

• Translate in coherent segments (sentences/paragraphs) rather than word-by-word — context improves quality and TM matching.
• Build up a translation memory by translating consistently within a project; the value compounds as reuse grows.
• For bulk or recurring translation, use a workflow loop that translates document chunks and joins the results.