Patois.

Localization Studio

The TMS you were going to build.

Localization Studio is a self-hosted, MT-first translation management system. You get the whole repo: host it, modify it, three years of updates included. It's yours.

No cloud. No subscriptions. No per-word fees.

Schedule a technical walkthrough → or email brent@patoislabs.com

A real run, uncut: TM match, translation, LQA, post-edit.

The problem

Your team has already had this conversation.

Somebody asked it in a meeting: "Why are we paying this much for a cloud TMS? Couldn't we build our own?" And the conversation died in that meeting.

Localization Studio is the missing starting point: a working, self-hosted TMS you could run internally tomorrow, and shape into whatever your team needs.

The build vs buy math →

The model

One arrangement. The whole repo. Yours.

One license, paid once. No subscriptions, no seats. You get the entire codebase: host it, connect your models, change anything. Three years of updates included, renewable after that, and every version you receive is yours forever. Your data never leaves your building, because there's no "our side" for it to go to.

And if Patois Labs disappears tomorrow, nothing turns off. You own working software, the source, and the right to keep running it. Forever.

The product

Three ideas the whole system is built on.

01

MT-first by design.

LLMs translate everything; humans handle the exceptions. Whole documents in one pass, not segment by segment. Automated LQA speaks MQM, and its findings feed the post-edit pass. Describe a workflow in a sentence; the assistant builds it.

02

Yours, entirely.

Runs entirely on a laptop. Point it at a frontier API, your approved models, or the GPU under your desk. Zero code changes. TM, glossaries, style guides, and an LLM-native memory layer feed every prompt. Fully offline on local models, or through the APIs your company has approved: that traffic follows your configuration, nobody else's.

03

Built like infrastructure.

A modular monolith with named ports: one process by default, your infrastructure by choice. Disable any feature and it's gone. Everything is an API first, with an MCP server on top, so your agents can drive the whole system. Deploys in minutes: Docker or plain Python, SQLite by default, no Kubernetes required.

Your data

Your translators' edits land in your lakehouse as training data.

Most TMS APIs give your ML team a keyhole. This one records every LLM call and every human correction, analysis-ready for your warehouse. It doesn't just do the work. It produces the dataset your next model trains on. Your content, your corrections, your dataset, under your existing vendor agreements.

The prompt tap: every request, every payload, and the export that follows.

Ready for your IT review

Self-hosted doesn't mean unserious.

SAML SSO · SCIM provisioning · Role-based access control · Full audit trail · Encrypted at rest · Web UI and CLI in feature parity

SSO and provisioning are interop-tested against a real Okta org; every action lands in a tamper-evident audit chain.

The full IT brief →

What's inside

A complete TMS. Built for how translation works now.

Translation memory.

Exact and fuzzy leverage, applied where it counts: inside the prompt. Standard TMX imports on day one.

Glossaries and style guides.

Enforced in the translation, not displayed beside it. Standard TBX imports too.

LLMemory retrieval.

RAG across everything you feed it: TM, terminology, style guides, project notes.

Vision context.

Screenshots ride along with strings, so the model sees the UI it is translating.

Documents and strings.

One-off document jobs or continuous, key-based localization. Both first-class.

Whole-document translation.

Full context in one pass. Segment mode when you want it.

Automated LQA.

MQM findings from an LLM reviewer, fed straight into the post-edit pass.

Workflows in a sentence.

Describe the pipeline you want; the assistant assembles it; you hit run.

The exception slice.

A real CAT editor for the strings that need a human.

API and MCP, day one.

Every feature is an API first, with an MCP server on top. Web and CLI in parity.

Connectors

Plugged into where your content lives.

Content flows in from your systems and comes back translated: Jira Airtable HubSpot WordPress Shopify Contentful and a catalog that keeps growing. Every connector ships as source, like everything else: adapt it to your setup.

Don't see your connector? Request it and I'll reach out.

Request a connector

Goes straight to brent@patoislabs.com. I read every one. Used only to reply to you, never shared. Privacy.

How it works

One pass through the pipeline.

Content comes in, machine translation does the bulk, quality and post-editing run automatically, and only the exception slice reaches a human. Then it ships, with the whole run recorded.

Localization Studio translation pipeline Content flows left to right through seven stages: content in (docs, strings, connectors); segment or translate the whole document; LLM translation, fed by translation memory, glossaries, style guides and LLMemory; automated LQA in MQM; LLM post-edit using the LQA findings; human review of the exception slice only; and delivery to files, strings, connectors and your lakehouse. TM · glossaries style guides · LLMemory Content in docs · strings connectors Segment or the whole document LLM translation Automated LQA MQM LLM post-edit with LQA findings Human review exception slice only Delivery files · strings connectors · lakehouse

Every request and response is captured whole. Your ML team finally gets all of it.

Having the build-vs-buy conversation?

Your hardware. Your models. Your data.

Let's talk. A 30 minute technical walkthrough: see it run, ask anything, decide if it fits your environment.

Built by a localization veteran of Intento, Lilt, LinkedIn, and Cricut.

Schedule a technical walkthrough →

or email brent@patoislabs.com