Score
An instant quality measurement from 0 to 100. Consistent and formula-based, so quality becomes comparable across documents and vendors.
Legal translation,
measured.
Lexmetra measures the quality of legal translations with AI calibrated by sworn translators, and turns that measurement into a complete review and certification workflow for enterprises.
Enterprise software, measuring real legal caseloads every day.
How it works
Upload. Measure. Act.
- 1
Upload
The source document and its translation. PDF or Word, from a two-page contract to a court file of hundreds of pages.
- 2
Measure
An instant Score from 0 to 100: a severity-weighted formula over a deep analysis of the full document pair, grounded in your own translation memories.
- 3
Act
Unlock the detailed Check report, request expert Human Review, and add legal certification when the document must stand before authorities.
Why it matters
In legal language, a mistranslation
is a liability.
Contracts, statutes, filings, evidence: meaning has to survive the crossing between languages, and someone has to be able to prove that it did. Lexmetra makes that proof measurable.
One funnel
From score to certified document
Check
The complete findings report: every issue with severity, category, excerpt and suggested fix. Bilingual, downloadable as Word.
Human Review
An expert translator corrects the document, assisted by the report, through a managed review queue with delivery tracking.
Legal certification
For documents that must stand before authorities: the deliverable carries the certifying translator's seal.
Each step builds on the last, and human review is part of every deliverable tier. The AI proposes, the expert owns the result.
For language service providers
Your firm's platform,
under your brand
Lexmetra brings the AI, the workflow and the infrastructure. You bring your expert translators and your client relationships: white-label branding, your client companies as sub-tenants, your prices, and every analysis grounded in your own translation memories.
The engine
Frontier models, on your terms
Lexmetra is model-agnostic by design: it runs on the strongest large language models available, closed and open-weights alike. Which one does the work follows your preference, not ours: a provider your organisation already trusts, a commercial agreement you hold, or open weights under your own governance. The platform adapts to it, not the other way around.
- Anthropic Claude model family
- OpenAI GPT model family
- Moonshot AI Kimi open-weight models
- Google DeepMind Gemini model family
- Meta AI Llama open-weight models
- xAI Grok model family
- Mistral AI Open and commercial models
- DeepSeek Open-weight reasoning models
- Alibaba Qwen model family
- Amazon Nova model family
- Cohere Command model family
- NVIDIA Nemotron open-weight models
No vendor lock-in: your model preference is configuration, not a renegotiation.
Anonymization
Identities out. Meaning intact.
Your documents carry your clients' identities. Ours is the technology that removes them: before AI, before sharing, before risk. Built into the platform, and available as a standalone service for document sets, memories and corpora.
Security & trust
Single-tenant by design
Isolated deployments
Each customer runs on its own instance, database and search index, in the Azure region it requires. No shared instance.
Corporate SSO
Microsoft and Google sign-in with corporate-domain gating and role-based access control.
Retention you control
Documents auto-delete on the schedule you set. Scores and accounting survive, content does not.
Never trains models
Client documents are processed for analysis only. They are never used to train models. DPA as standard.
Built on Microsoft Azure enterprise infrastructure.
The full security posture →Next step
See it on your own documents
A demo runs on a real document pair from your world: your document types, your volumes, your workflow. No canned samples.