Translation memory, terminology, and MQM review in one editor. Bring your own AI keys or pay only for what you use. Linguists keep every final call.
We're building locAGI to be the cloud-hosted, audit-ready translation platform the largest localisation programmes need but can't get from today's CAT/TMS vendors. A single managed stack for terminology governance, AI-assisted translation, MQM scoring, and post-processing — with linguists in charge at every step, transparent usage-based billing, and a free tier that actually does real work.
The full CAT environment: source/target editor, TM with concordance, terminology, comments, MQM review. Native — not bolted on.
Scheduled for Q2 2027. We’re stabilising the surface and completing the security audit. Until then locAGI is in internal use — join the waiting list ›
The honest version: substantial pieces are still in flight. This is where we stand — not where we’d like to look.
Translation memory, terminology, AI post-editing and MQM review in a single editor — so a job never leaves the platform to get finished, and every change is attributable.
A project termbase and a running glossary constrain every AI call, so a term decided in batch 2 still holds in batch 40. Translation memory is searchable across projects, with concordance and pretranslate.
Bring your own keys or let locAGI meter the tokens. Fuzzy bands decide how hard the model works — a 97% match gets a light touch, a low match gets a full post-edit — so you are not paying to retranslate what already matches.
Score against MQM Core or LISA QA Model 3.1, or free-text categories. Findings carry severity, comment and a suggested revision, and travel back into the bilingual file as native CAT comments.
Round-trip the bilingual formats your clients and vendors send, and pull everything back out again. No proprietary container, no lock-in.
Localisation work carries client confidentiality. The platform is built so the sensitive parts stay under your control rather than ours.
Engineering decisions that make locAGI safe to hand to your compliance team and trust with regulated content — your text goes to the AI provider you choose, never through anyone else's model.
Long files don't drift. The AI engine carries your project termbase + a running glossary harvested from earlier segments so terminology and tone stay consistent across a 10,000-segment document — not just within a single paragraph. Partial failures retry per-segment so nothing is dropped silently, and every batch runs on low-temperature prompts, logged so a run can be traced.
Bring your own keys (BYOK): your OpenAI / Anthropic / Google API keys, the call goes straight to the provider, the card you have on file with them is the one charged. Or let locAGI provision the tokens for you on a pay-as-you-go meter — one bill, no procurement work, mix-and-match across users in the same workspace. Your compliance team — not us — decides which providers are allowed for which data.
Prompt templates carry your style guide and instructions into every AI pass, with the termbases and translation memories you link to them. Point locAGI at any OpenAI-compatible endpoint, a private model included — your compliance team decides which providers see which data.
Multiple linguists in the same project at the same time — presence indicators show who's where, autosave as you work, live status broadcasts so a reviewer sees your translation the moment you confirm it. Comment threads with @mentions for clarification questions. Freelancers outside your workspace get a secure vendor link: they translate or review their own files in the browser, with no account and nothing to install. memoQ packages go in and come back as return packages.
Read projects, segments and QA findings, save translations, pretranslate from your translation memories, run QA and export the translated file — over HTTP, with API keys. Wire locAGI into your TMS, your CI, or your project-management tool.
locAGI ingests source documents directly — Word, Excel, Markdown, HTML, JSON, YAML, XML, CSV, plain text. No separate file-prep step, no "first run it through OpenXLIFF yourself." For bilingual workflows you also get the full CAT-tool interop stack: XLIFF 2.0, MQXLIFF (memoQ), SDLXLIFF (Trados), TMX (any TM), TBX (any termbase). Round-trip preserves every tag, every placeholder, every locked segment.
Upload one source file, add as many target languages as you need — locAGI runs MQM review across all of them in one workspace. One translation memory shared across the project, one termbase per project, one MQM scorecard per language. Reviewers see all locales side-by-side; project managers see one rollup. No more juggling six CAT-tool installs to QA six languages, no per-pair licensing math.
Hit one button and the AI rewrites the target using your project glossary, style guide, and the surrounding context. Need an explanation rather than a fix? Open the segment chat and ask the AI a question in plain English — "why did you pick this term?", "what's the formal register equivalent?", "is this idiomatic?". Each segment also gets a predicted next translation you can accept, edit, or ignore. Everything respects your BYOK or pay-as-you-go choice from above.
From the public release, sign up free and invite your whole linguistic team — internal staff and contracted freelancers — without paying per head. AI usage runs on tokens we provision (pay-as-you-go) or on your own OpenAI / Anthropic / Google API keys (BYOK). Storage is generous out of the gate; you only pay when you exceed it. Transparent meter, no surprise overage fees.
Traditional CAT tools charge per-seat subscriptions or thousand-euro perpetual licenses, then put AI behind a separate plugin. locAGI charges €0 per seat — ever. This is the pricing from the public release in Q2 2027; today locAGI is in internal use.
locAGI is an integrated cloud CAT platform — translation memory, terminology, MQM review, AI-assisted translation — free to sign up from the public release, with pay-as-you-go AI or bring-your-own-key. Your projects, your linguists, your linguistic standards — locAGI runs in the background.