Architecture comparison
A deployment and control comparison for law firms evaluating where sensitive matter data is processed, governed and retained.
Shared cloud legal AI
Your documents are processed on the vendor's infrastructure.
Your firm uses a multi-tenant, vendor-operated cloud service shared with other customers.
Your security boundary extends to infrastructure operated outside the firm.
A compromise of the vendor environment can put customer work product in scope.
General assistants
Not built around privilege, matter scope or the matter as the operating unit.
Legal source verification and matter provenance are not the default interaction model.
Data handling depends on the assistant, plan, settings and vendor policy.
The environment is designed for general assistance rather than governed institutional legal work.
Sanctum Lex
Inference runs inside your perimeter, on your hardware.
One firm per node. Nobody shares your enclave.
Zero external model egress by architecture.
If we are breached, your private-perimeter matter documents were never in our vendor environment.
Matter intelligence, legal research, drafting, review, knowledge and governed workflows remain attached to the same matter context.
Source-grounded responses return the passage, page and matter reference for lawyer review.
SSO, SCIM, ethical walls, matter permissions, audit, retention and model policy operate as firm-level controls.
Institutional knowledge and approved methods remain governed alongside the matter.
Opposing Counsel and Judicial Review can independently pressure-test the same position without rebuilding context.
Hearing Room and Witness Room carry the same matter into preparation and rehearsal.
The firm controls the deployment boundary instead of extending privilege into a vendor-operated AI environment.
Architecture review
Give your lawyers a persistent intelligence environment for the matter, with the controls your firm requires and an evidence trail they can review. These are the commitments to establish together before deployment.
Your firm can bring advanced legal intelligence to sensitive work within an approved processing boundary. In a private-perimeter deployment, matter documents, prompts, outputs and model inference stay within that environment. The deployment record identifies the hosting location, storage boundary and inference endpoints. Any approved external model or research service is documented separately, including its permitted data path and any cross-border transfer.
Your firm sets the rules for how Sanctum operates. In a private-perimeter deployment, the firm controls the hosting infrastructure and network isolation. Sanctum applies configured identity, provisioning, matter permissions, ethical walls and approved model routes, while retention requirements follow deployment policy and the client agreement. Infrastructure restrictions and application controls are reviewed together, giving the firm a documented view of the active configuration and verified controls.
Your lawyers can inspect the evidence behind source-backed analysis and work product. Sanctum keeps document, page and passage references connected to the matter throughout analysis, drafting and review, so the lawyer can open a cited source and examine its provenance. Authority results retain links to their underlying sources and their verification status. Evidence gaps and incomplete verification remain visible, supporting informed professional judgment.
Bring representative documents and one workflow your lawyers care about. We will show the legal work and the private-perimeter deployment boundary in the same session.