Literess now looks things up online while she answers and lists what she read, and long conversations compact themselves instead of filling up. Glossaries and style guides gained a complete version per language, and a document that already exists can grow new languages over the API.
Everything the editor does is now reachable over a public API with scoped keys, and behind an MCP endpoint that Claude Code, Claude Desktop, or Cursor connect to with a browser sign-in. Translation changed underneath too: house style rules apply to every run without a style guide, and the starting-point templates translate on Pro.
Software strings and email templates now import as first-class documents — CSV, gettext PO and XLIFF 1.2 — with placeholder tokens protected end to end and a grid built for reviewing them. Glossaries and style guides got their own library pages, multiple target languages, and a review step before anything generated lands.
A new suite of free, browser-based converters for the file formats localization work runs on — XLIFF ⇄ Excel, XLIFF ⇄ CSV, and a TMX cleaner — with no sign-up and nothing uploaded. Plus glossaries and style guides you can build from a whole document, and a batch of translation-memory and reliability fixes.
Running a workflow across all of a document’s languages is now one action with one progress card — a lane per language, one notification each, one summary at the end. Reviews collect in a Reviews ready list instead of interrupting you, and Literess can now check in on its own.
The app and marketing site are now available in ten languages — French, Spanish, Italian, Czech, Polish, and Turkish joined this week — and the Journal is localized too. Plus a “Translate + Proofread” editor action, a scope picker for runs, and AI-sync review for glossaries and style guides.
Teams are now seat-based with a shared word pool, so a whole team bills on one invoice. Plan and pack prices also moved — Starter €29/mo, Pro €79/mo — while words, the Free plan, and per-word rates stay the same. Plus a simpler Proofread, direct PDF export, and a batch of editor fixes.
Transept now remembers the sentences you’ve already translated — and does what a classic CAT tool can’t: it feeds your past wording to the model as reference, so the AI reuses your phrasing instead of inventing new. Free on every plan.
The editor moved to a resizable, panel-based layout with a calmer visual refresh — and Transept now works on a phone, so you can review and run translations on the go.
Describe the run you want and Literess assembles the whole workflow — or edits the one already open in the builder — and shows her proposal on the canvas before you save. A new in-app help menu adds a front door to support.
Runs now show their cost up front and never spend more words than you have — your balance can no longer go negative, and a document that doesn’t fully fit translates as much as it can from the top.
Take your translations and terminology into any CAT or translation-memory tool — documents now export as TMX, XLIFF, or CSV, and glossaries as CSV or TBX.
Literess stopped being an assistant that only explains things. Now she sets up glossaries, runs workflows, moves work along your pipeline, and drives the UI for you — after you approve.
The business side of translation now lives in one place — a job board, your client book, projects, tasks, and stats. Built for freelancers and agencies.
A visual builder for chaining translate, proofread, improve, check, and export into one repeatable run — with live progress and a cancel button that actually stops the model.
Step through a workflow’s findings one at a time — accept, reject, or edit each — and start new documents from a template tuned to what you’re translating.
Glossaries gained structured per-term controls and an AI consistency check; styleguides became a structured editor with presets and a build-from-a-sample wizard.
A proofread that looks at the rendered page like a reader — for HTML documents and full-body emails, catching layout and visual issues text alone can’t.