
TranslationOS
TranslationOS is Translated's enterprise localization platform. It runs the publisher's Lara translation AI alongside professional linguists in a single workflow, giving teams real-time visibility over progress, quality, terminology and spend, plus API and connector integration.
What is TranslationOS?
TranslationOS is the enterprise localization platform published by Translated S.r.l., a Rome company whose About page states it was founded in 1999 by linguist Isabelle Andrieu and computer scientist Marco Trombetti and now employs around 230 people across Europe and the United States. The publisher calls it an adaptive AI service delivery platform for translation, and presents it as the entry point to its wider services.
Its organising idea is a move away from workflows built around translation memories towards workflows built around full content context. Within one system the platform coordinates Lara, Translated's proprietary translation AI, and human language professionals, applying the level of human review that the business context and the delivery contract call for. Lara has been reachable through the platform's API since July 2025. The publisher says it was trained on more than two billion professionally translated sentences, including revisions and discussions exchanged between translators and revisers, and that a technique it calls Trust Attention makes the model favour verified professional translations over unverified web content. Claimed coverage is more than 200 languages and locales, served by a network of 500,000 professionals across 40 specialist domains.
What the platform adds on top is observability. Teams follow progress, quality, terminology and spend as work moves from draft to approved, see which linguist changed what, and read quality as Errors Per Thousand words measured against a threshold fixed in the contract. Access is governed by single sign-on and four roles, Admin, Localization Manager, Viewer and Finance, the last of which sees spend and invoices.
Content reaches the platform three ways: eighteen ready-made connectors covering content systems, translation management systems, code repositories, marketing tools and design files; custom connectors built for in-house systems; or the documented TranslationOS API, which developers drive from a CI/CD pipeline and can exercise against a sandbox environment. The platform handles 71 file formats and runs with or without an existing translation management system.
What it does
- Send content for translation from a single entry point, with or without an existing translation management system
- Coordinate the Lara translation AI and professional linguists in one workflow, with human review set by business context and contract
- Track progress, quality, terminology and spend in real time, from request through to approved delivery
- See how the selected linguists changed meaning, tone and context on each job
- Feed every approved translation back into the AI to raise first-pass acceptance over time
- Connect content systems, translation management systems, code repositories, marketing tools and design files through ready-made connectors
- Drive localization programmatically from a CI/CD pipeline through the documented TranslationOS API
When to use TranslationOS / When not to
A quick filter to help you decide if TranslationOS is the right fit.
When to use TranslationOS
- Enterprise localization managers industrialising a multilingual content programme across many markets
- Product and engineering teams that want continuous localization inside their CI/CD pipelines, through the API or the GitHub connector
- Marketing and lifecycle teams localising campaigns held in Marketo Engage, Iterable or SendGrid
- Support and knowledge-base owners translating help content stored in Zendesk or Salesforce Knowledge
- Finance and operations leads who need localization spend, invoices and delivery risk in a single view
When not to use TranslationOS
- Individuals and small businesses needing a one-off document translation, whom the publisher routes to its public instant-quote engine instead
- Buyers who need a published rate card, since no price is disclosed for the platform and there is no self-service sign-up
- Teams that want to test a free plan or a free trial before committing, as neither is offered
- Anyone expecting a mobile application: this is a web platform with an API, with no iOS or Android release
- Organisations that require a signed data processing agreement and a published subprocessor list before onboarding, as neither is available on the site
How to use TranslationOS
A typical end-to-end flow, from setup to results.
- Request a demonstration from the product page form, or reach the team through the contact page
- Agree the approach with the publisher: build a localization programme from scratch, or introduce AI gradually into an existing one through a pilot
- Sign in to the platform at os.translated.com, using single sign-on or email and password credentials
- Have an administrator add users through the identity provider or the account manager, and assign the Admin, Localization Manager, Viewer and Finance roles
- Connect the content sources by choosing a connector, following its authentication flow, then selecting which content to translate
- Alternatively, call the TranslationOS API directly from a CI/CD pipeline, testing against the sandbox before moving to production
- Upload translation memories, glossaries, context and custom instructions so the AI follows house terminology
- Send content and let the workflow route it between the AI and the linguists at the agreed review level
- Follow progress, quality scores, terminology and spend on the dashboards, and inspect the linguists page when a score drifts
- Approve deliveries, each approval feeding back into the AI for the following projects
Pros & Cons
Pros
- The API is publicly documented and versioned, and comes with a sandbox environment, which is uncommon among enterprise localization platforms
- Eighteen named official connectors span content systems, translation management systems, marketing tools, code repositories and design files, with custom connectors offered
- The platform runs with or without a translation management system, so it does not force a replacement of existing tooling
- End-to-end traceability is the core promise: who did what, when, and at what cost, from request to approval
- The publisher has operated since 1999, holds ISO 9001:2015 and ISO 17100:2015 certification, and names verifiable enterprise references
- Commitments are quantified: a contractual quality threshold expressed in errors per thousand words, and a refund when delivery is late beyond a third of the agreed time
- A public vulnerability disclosure programme with safe harbour explicitly covers the platform and API hosts
Cons
- No price is published for the platform: no rate card, no tier, no order of magnitude, with access running through a demonstration and a contract
- The rates the publisher does display, around USD 0.10 to 0.11 per source word, cover its general translation services and not this platform, which that page never mentions
- There is no free plan, no free trial and no self-service onboarding
- The privacy policy has not been updated since 25 May 2018 and still relies on the Privacy Shield for non-EU transfers, a mechanism invalidated in July 2020
- No data processing agreement and no subprocessor list are published, although the platform handles enterprise content
- No documented way exists to exclude your content from model training, while the publisher states that every approved translation improves its AI
- No information security certification such as SOC 2 or ISO 27001 is claimed; the certifications shown cover quality management and translation services
Pricing & Plans
No public price exists for TranslationOS. The publisher discloses no rate card, no tier and no entry-level amount for the platform: the only route shown is a demonstration request followed by a contract, with the API page describing monthly invoices and a translation quote based on the agreed process and contractual service level. There is no permanent free plan and no free trial. The rates that Translated does publish elsewhere, around USD 0.10 to 0.11 per source word and roughly USD 25 for a standard page, apply to its general translation services and not to this platform, which is not mentioned on that page. Buyers should expect a negotiated, volume-based agreement.
Data, GDPR & hosting
A consolidated view of how TranslationOS handles your data.
GDPR overview
Compliance is claimed explicitly: the privacy policy opens by invoking Regulation (EU) 2016/679 and Italian data protection law. The controller is Translated S.r.l., Via Indonesia 23, Rome, and a Data Protection Officer is named, Simone Giovanni Perone, reachable at dpo@translated.com. Processing rests on consent for site visitors and on Article 6 for client data. Listed rights cover access, origin, rectification, erasure, anonymisation, blocking, portability and complaint to the supervisory authority. Two gaps stand out. The policy was published and last updated on 25 May 2018 and still justifies non-EU transfers under Privacy Shield decision 1250/2016, which Schrems II invalidated in July 2020. And no data processing agreement or subprocessor list is published anywhere on the site. No Article 27 representative is designated, which is consistent, since the publisher is established in the European Union.
Who owns the data?
Translated's sales conditions are silent on who owns the delivered translation: no assignment and no licence clause appears. The only ownership statement is a negative one, namely that until the invoice is paid any partial or total reproduction of the translation is deemed illegal, with the publisher reserving the right to claim payment and copyright compensation. On confidentiality the commitment is broader. Source texts, the nature of the work performed, personal data submitted for translation and any information transmitted by the client are held under the strictest confidentiality. Disclosure to third parties requires the client's prior written consent, except to authorised employees and to authorised subcontractors whose intervention requires it.
Reuse rights
The sales conditions give the client no explicit licence over the delivered translation, and set out no explicit restriction on reusing it either, beyond one condition: reproduction remains illegal while the invoice is unpaid. In practice a client who has paid uses the translation as their own content, but nothing in writing confirms it, which matters for regulated or high-value material. Quality claims are tightly framed. Complaints are admissible only by registered letter with return receipt within seven days of delivery, after which the translation is considered correct, and liability is capped at the invoice amount. Disputes fall to the courts of Rome and only written agreements are taken into account. Inside the platform the client supplies translation memories, glossaries, context and instructions, and approved translations are fed back into the publisher's AI.
Data retention & training
Hosting summary
The privacy policy states that data collected through the site is processed at the controller's headquarters and at the web host's data centre. One host is named as a processor: Hitachi CBT, described as located in the European Economic Area and operating to European standards. No specific country is identified, so the record carries a region rather than a country. Transfers outside the European Union are acknowledged towards Google, Facebook and Microsoft through social plugins and analytics, still justified under Privacy Shield decision 1250/2016, a basis invalidated in July 2020. Governing law is Italian and commercial disputes fall to the courts of Rome. The public site itself is served through Amazon CloudFront, so the resolved server address reflects an edge node rather than any storage location. The platform and its API run on separate hosts, both named in the vulnerability disclosure scope. No hosting information specific to this platform is published: the policy addresses the main website.
Things to keep in mind
Risks and trade-offs to weigh before adopting TranslationOS.
- Concentrating the whole multilingual pipeline on one vendor makes the platform a single point of dependency for every market a company serves
- Approved translations feed the publisher's AI with no documented opt-out, so potentially confidential enterprise content contributes to a model
- Commercial opacity works against the buyer: with a negotiated price, a contractual quality threshold and no public grid, there is nothing to benchmark against
- Automatic scores such as errors per thousand words and acceptance rate can become targets in themselves, nudging teams to wave through mediocre output to protect a metric
- Internal language skills can quietly erode if reviewers stop reading carefully and start trusting the first-pass acceptance rate
- The privacy policy has stood still since 2018 and still leans on an invalidated transfer mechanism, so a client data protection officer will need an up-to-date framework before sending personal data
- With no published subprocessor list, the chain of parties handling entrusted content is not visible from outside
Setup & Integrations
Technical difficulty
Setup is assisted rather than self-service: onboarding starts with a demonstration and an account manager. The simple route is a ready-made connector, where a team picks its system, follows an authentication flow and configures which content to translate. The technical route is the API, with API keys, a sandbox and nineteen endpoints, plus a software development kit and command line tools for pipeline integration. Single sign-on must be configured on the identity provider side. Nothing is installed locally. The publisher advertises a free migration plan and custom connectors for in-house systems.
Deployment
Integrations
Supported languages
Behind TranslationOS
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What is TranslationOS?
Who publishes it?
How much does it cost?
Is there a free plan or a free trial?
Is there an API?
Which tools does it connect to?
Do I need an existing translation management system?
Is my content used to train the AI?
Where is data hosted, and is a data processing agreement available?
Which certifications does the publisher hold?
Should you pick TranslationOS?
TranslationOS is a mature enterprise product rather than a tool an individual can pick up. Its centre of gravity is operational transparency: seeing who did what, at which quality, at what cost, as content moves from draft to approved. That promise is backed by things a buyer can verify from outside, which is what makes the platform interesting. The API is publicly documented and versioned, it exposes a sandbox, eighteen official connectors are named one by one, and the platform is explicitly designed to run with or without an existing translation management system, so it does not demand that a team throw away its tooling. Behind it sits a publisher that has been operating since 1999, holds ISO 9001 and ISO 17100 certification, and puts named enterprise references on the record.
The reservations are real and fall into two groups. Commercially, nothing is public: no rate card, no tier, no free plan, no trial, and no self-service entry. The rates Translated publishes belong to its general translation services and should not be read as the platform's price. Legally, the privacy policy has not moved since May 2018 and still cites the Privacy Shield for non-EU transfers, a mechanism struck down in 2020; no data processing agreement or subprocessor list is published; and although the publisher states that every approved translation improves its AI, no opt-out is documented. For an enterprise buyer with a procurement process, those are contract-stage questions rather than blockers, but they should be asked. Note too that the figures the site quotes for clients, languages and translator network size differ from one page to another, so treat them as publisher claims rather than settled facts.
- Choosing a selection results in a full page refresh.
- Opens in a new window.