AI retrieval
Scoped by tenant, case, document accessibility, hashed metadata, post-validation, and stale vector rejection on the hardened path.
Trust Center
The controls around one central system for sensitive casework.
A central trust portal for teams evaluating Caseflow before bringing legal, investigative, inheritance, and other confidential case records out of fragmented tools and into one workspace.
Enterprise review hub
Caseflow trust materials support procurement, security review, privacy assessment, and AI governance discussions with clear language about how the system handles sensitive casework.
- Tenant-scoped access controls and role-aware workspace permissions.
- Permission-aware AI retrieval constrained by accessible case, task, and document records.
- Document-scoped authorization with short-lived access links in migrated document flows.
- Reviewable AI actions with confirmation workflows for supported write operations.
- Clear trust information written for product review, onboarding, and support conversations.
Trust resources
Security
The access, authorization, AI safeguards, and auditability controls around your central case record.
AI privacy
Controlled AI processing, permission-aware retrieval, vector filtering, human review workflows, retention, deletion, and known limitations.
Privacy policy
How workspace content, account data, collaboration records, support communications, billing metadata, and AI-assisted processing are handled.
Terms
Terms for professional account use, customer content, collaboration permissions, generated outputs, and third-party services.
Cookies and browser storage
Cookie, local storage, service worker, installable app, localization, preference, and authentication-state disclosures.
Support
Support for onboarding, moving into the workspace, account access, workflow questions, and privacy or AI handling concerns.
What is strong today
The strongest trust areas are the main AI retrieval path, tenant/case/document scoping, permission-aware context rebuilding, and confirmation-gated AI write actions.
Document access
Migrated document views request permission-checked short-lived access links instead of reusing stored token URLs.
Human review
AI output is positioned as assistive and reviewable, with professional decisions left to users.
What remains transparent
Some legacy or out-of-scope surfaces still require migration plans, and public pages avoid claiming that every URL, attachment, or processing path has identical controls.
- Legacy URL-only records remain supported for compatibility.
- Pattern redaction reduces exposure but is not guaranteed anonymization.
- Provider processing and retention commitments depend on approved configuration and contract terms.