Control
Translation memory that remembers your decisions
No cardFirst 5 pages freeYour work never trains AI
Literess, the AI editor in Transept
You: Make Grandma grumpier.
Literess: Done. She doesn’t just say it anymore. She grumbles it. sagte → raunzte
Ask in plain words. You approve every change. Meet Literess →
How do you translate it once and reuse it everywhere?
In a classic CAT tool, you maintain the translation memory and copy from it by hand. Transept builds it as you work and feeds it back into the model, so the next translation already knows your past decisions. Generic AI tools can’t do that.
It remembers as you go
Every block you or your team approves goes into memory automatically. There’s no separate database to maintain and no export step: as soon as a translation is active, it’s reusable.
Matches appear as you translate
Select a block and the Memory panel shows the closest past translations. Matches come from meaning as well as wording: a semantic search finds segments that say the same thing in different words, alongside exact and fuzzy text matches. A percentage badge shows how close each match is, and one click copies the translation or inserts it into the block.
The AI translates with your memory
Your matches go into the prompt as reference, so translation, proofreading, and Find reuse your established wording instead of inventing a new phrasing each time. A glossary can’t do that, and a raw LLM won’t.
Bring your own, search across everything
Import an existing memory as TMX or XLIFF and it’s searchable from day one. Search one document, a project, or the whole team, and include past comments and chat alongside the translations.
Every translation tool has a translation memory. Trados and memoQ built the category, and you can get one free in an open-source editor. So the real question is what the memory does. The classic answer is retrieval: have we translated this exact segment before? Pull the match and fill the gap. Transept’s answer is context: what does this project already know, and how much of it belongs in the sentence you’re working on? When you approve a translation, Transept remembers it automatically, shows it to you as you work, and hands it to the model as reference, so the next draft reuses your wording instead of starting cold. There’s no database to build, no concordance to maintain, and no copying and pasting from a side panel. You don’t query this memory; it works alongside you while you translate.
Where does reuse pay for itself?
Repetitive documents
Manuals, contracts, release notes, and product catalogs, where the same sentence keeps coming back. Insert the past translation from memory at no cost, or let the AI draft with it as reference so the wording stays the same.
Consistency across a library
A help center or doc set where every warning, step, and piece of boilerplate has to read the same in every article and every language.
Memory that remembers the why
The rejected alternatives and the comments behind a choice come back with the final wording, so the model reuses your judgment as well as your words.
4× cheaper model, matched
Give a model the decisions behind past translations, not just their final wording, and it stops guessing. With enough of that context, a model costing a quarter as much matched a premium one on our own benchmarks.
The memory betOn matching alone, the field has hit a ceiling. Trados and memoQ are the best of the classic approach: exact and fuzzy matches, and at the top, context matches (memoQ’s 101%: a segment plus its neighbors). It’s a database you query, and it has been perfected. What it can’t do is find a sentence that means the same thing in different words. Transept combines three kinds of matching (exact, fuzzy, and semantic, which works on meaning) into one ranked list, so a paraphrase turns up alongside a near-duplicate. Semantic search on its own is risky, because it will accept a near-twin like “ileum” for “ilium.” Combining it with exact and fuzzy matching keeps it precise.
The bigger difference is where the memory lives. In Phrase, Smartcat, and the rest, it’s an organized database beside the editor: you query it, you fill the segment, and an MT engine is chosen for the draft. Lilt goes further and fine-tunes a model on your past work, but the result is model weights you can’t inspect, not retrieval you can. Transept keeps the memory in the work itself: it feeds the draft and the QA pass, you can search it across past comments, chats, and whole documents as well as segment pairs, and it moves with the project instead of sitting in a box you have to remember to open. (For string-level software localization, with a memory keyed to message IDs and wired into CI/CD, a tool like Lokalise is still the better pick. Transept is built for documents and content.)
— Don’t take our word for it —
The editor, in miniature
A working slice of the real app: Literess, the glossary, the styleguide, workflows, and translation memory are all live. Click around.
Simple pricing
Every plan comes with monthly credits. You spend them only on AI work, at the price each action shows before you click. No sales call and no contract: start free and add a credit pack whenever you need more.
One credit covers one source word in Standard mode. Pro mode uses a more capable model and costs 3 credits per source word. Chinese and Japanese source text is counted per character. Manual edits, exports, comments and translation memory are free: you spend credits only when AI translates, rewrites or checks your text, or reads a document to build a glossary or styleguide. Every paid action shows its price on the button before you click. See the price list.
Free
Always freeFor your first projects
Unlimited users
Translated with Pro, proofread with Standard. About 11 on Standard alone.
- Pro mode: a more capable model for nuance, 3 credits per source word
- Standard quality: fast, accurate, faithful
- Translation memory: your past translations, reused to keep new ones consistent
- Workflows: chain translation, proofreading, and checks into one run
- Glossary, styleguide, QA, import / export: all included
- Notion, Drive, TMX / XLIFF: import, export, and bring your own memory
- 68 languages: from Spanish to Tagalog
- Batch translation of whole documents
- 10 lifetime messages with Literess: your AI literary editor, who remembers your projects and preferences
- No card required
Starter
For regular client work
Unlimited users
- Literess: your AI literary editor, who drafts, proofreads, and sets things up for you
- API, MCP & webhooks: automate runs, or let AI assistants work in your account
- Teams: invite anyone for free; add €29 seats to fund a shared credit pool (60,000 credits a month each)
- Top up anytime: pack credits never expire
Pro
For higher volumeFor agencies & large catalogs
Unlimited users
- Over 3× Starter’s credits: headroom for batch work and full catalogs
- Priority support: a direct line and faster replies
- Predictable bill: most months need no extra credit packs
Prices are in euros and exclude VAT. If VAT applies to your purchase, checkout shows it before you pay, and a business with a valid EU VAT ID pays the net price under reverse charge. You can buy credit packs on any plan, including Free: 30,000 credits for €19, 100,000 for €55 or 300,000 for €149. Pack credits never expire. Creating a team and inviting people is free. To share credits, add two or more seats: each seat adds its plan’s monthly credits to the team’s pool, all on one invoice. Cancel anytime and keep your plan until the end of the period you’ve paid for. We give refunds for faults on our side that we can’t fix and for billing mistakes, and EU consumers can withdraw within 14 days and get back the unused part.
FAQ
- Yes. It’s part of every plan, including Free, with no add-on to buy and no separate database to manage. Every translation you approve goes into your memory automatically.
- A glossary pins individual terms, such as a name, a product word, or a phrase. Translation memory matches whole segments: sentences and paragraphs you’ve already translated. The glossary keeps words consistent, and the memory saves you from translating the same sentence twice. Most professionals use both.
- Yes. Export it as TMX or XLIFF and upload it. Your existing source and target pairs are searchable right away and feed the AI like the rest of your memory.
- No. Copying or inserting a match from your memory costs nothing, and matches in the AI’s prompt don’t add to the price. Credits are spent when AI works on your text (translating, rewriting, checking), at the price shown before you click: a block the AI translates bills its source words at the normal rate, with or without memory behind it.
- Trados and memoQ offer the best of classic CAT memory: a segment database with exact, fuzzy, and context (101%) matches. Transept differs in two ways. The first is matching: they stop at text similarity, while Transept adds semantic search, so it finds a sentence that means the same thing even when it’s worded differently. The second is where the memory lives: theirs is a panel you query beside the editor, while Transept’s builds itself from the work you approve and feeds the AI draft and the QA pass.
- Phrase and Smartcat are strong platforms that organize memory across projects and clients. Phrase picks a machine translation engine for you, scores each segment, and can pass fuzzy matches from your memory to its own model. Lilt fine-tunes a model on your past work instead. That’s powerful, but the result is model weights, not retrieval you can inspect or import. In Transept, the memory is the context the AI translates in, built from exact, fuzzy, and semantic matches you can see.
- For document and content work, usually not: translation, memory, glossary, QA, and export are all in one place in Transept. For string-level software localization (a TM keyed to message IDs and wired into CI/CD), pair Transept with a tool like Lokalise. Import and export use TMX and XLIFF, so nothing is locked in either way.
- It’s translation memory adapted for large language models. Instead of only retrieving a past segment for you to paste in, Transept passes your matches into the translation prompt as reference, so the machine translation reuses your established wording. It’s the same idea as a classic TM, applied to the way the AI actually writes the text.
Keep reading around this capability
- FeatureA glossary the AI actually follows→Character names, product terms, brand phrases, client vocabulary. Build a glossary once (by hand, from a file, or from a sample document)…
- FeatureA styleguide that steers every regeneration→Tone, register, rules, formatting conventions. Write a styleguide yourself or have Transept generate one from a page of finished work. It…
- CompareTransept vs Phrase self-serve vs enterprise→Phrase is a powerful enterprise localization platform with a deep TMS, vendor management, integrations, and LQA. Transept is a different…
- CompareTransept vs Lokalise for content work→Lokalise is one of the strongest enterprise localization platforms, built for product strings, continuous localization, and developer…
- FeatureBatch translation without the spreadsheet→Taking a document into 12 languages usually means a spreadsheet of files, a project manager, and a week of follow-up. With batch…
Built for work you answer for
Private by default
Your text is never used to train AI, not by us and not by any provider we use, and it stays out of our logs. Delete your account and your text is deleted with it.
How we handle dataSecurity you can check
Hosted in Germany, in Hetzner’s ISO 27001-certified data center. Every sub-processor is listed, and a DPA is one email away.
All sub-processorsGive it context
Glossaries, styleguides, translation memory and per-string notes go with every request. You set them up once per project.
GlossariesYou’re the human in the loop
The AI drafts; you decide. Review every change side by side, keep or revert it, and send a review link when a client needs to look.
Review linksAPI, MCP and webhooks
The full API on Starter and Pro. Connect Claude or another AI assistant over MCP, or start runs from your own tools.
Developer guideSomething bigger?
Large catalogs, custom workflows, a DPA for procurement. Transept stays self-serve; write to us for the rest.
support@transept.aiFeatured on
