Isolated deployments
Every customer gets a physically isolated single-tenant deployment: its own instance, its own database, its own search index. Nothing is shared with any other customer.
Security & trust
Single-tenant
by design
Legal documents are the most sensitive text an organisation produces. Lexmetra's architecture starts from that fact: no shared instance, no shared data, no training on your documents.
Data residency by design
Each deployment is placed in the Azure region and data zone your organisation requires. Residency is an architecture decision, not a support ticket.
Corporate SSO
Sign-in through Microsoft Entra: Microsoft and Google corporate accounts, with corporate-domain gating and personal-domain blocking. Role-based access control keeps vendor and customer sides strictly separated.
Partitioned memories
Translation memories are partitioned per client company, and retrieval is scoped to match: one client's data is never used to process another client's documents.
Retention you control
Document retention is configurable per client company, with automatic deletion on schedule. Scores and accounting survive for your records; the content itself does not.
Never trains models
Client documents are processed for analysis only. They are never used to train models. A data processing agreement is standard with every deployment.
De-identified by default
Every translation-memory asset is de-identified by Lexmetra's own anonymization technology before it enters the AI layer. Your clients' identities never ride along with your firm's linguistic assets.
Built on Microsoft Azure enterprise infrastructure.
The anonymization technology →Questions
Ask us the hard ones
Security reviews are welcome. Bring your questionnaire, your DPA and your infosec team; we will walk through the architecture deployment by deployment.