Skip to content
Valentry

Trust & governance

Control should be visible in the workflow.

Valentry is designed for regulated insurance work where access, versions, decisions, evidence and report integrity must remain clear long after the task is complete.
Discuss governanceReview the controls
Access
Role · organisation · assignment
Audit
Important actions recorded
Versions
Analysis and report integrity
AI review
Human decision remains visible

Product control model

Security is not a badge row. It is how work is allowed to happen.

Control through permissions, versioning, audit events, governed review and protected report access. Deployment-specific architecture and contractual commitments are confirmed during implementation.

  1. 1Access

    Role-based visibility

    A user should access a case only when ownership, organisation, assignment or administrative permission allows it.

  2. 2Traceability

    Immutable audit events

    Important login, case, file, analysis, review, report and administrative actions should remain recorded and attributable.

  3. 3Versioning

    No silent overwrites

    Completed analyses create new versions. Final reports remain immutable and tied to the exact analysis version used.

  4. 4Reports

    Controlled downloads

    Users should be authenticated, authorised for the case and limited to reports that are available and not revoked.

  5. 5Data

    Protected client information

    User credentials, personal information, insurance documents, analysis results, reports and organisation data are protected by design.

  6. 6AI

    Human review remains explicit

    Reviewers can approve, reject, request information and record quality or compliance concerns against the relevant version.

Responsibility matrix

The right control for the right role.

A simplified view of the separation between standard users, reviewers, organisation administrators and platform administrators.
CapabilityStandard userReviewerOrganisation adminPlatform admin
Create and manage permitted casesYesYesOrganisation scopePermitted scope
Upload and categorise documentsYesYesOrganisation scopePermitted scope
Approve or reject analysisNot by defaultYesRole dependentRole dependent
Mark reports finalIf permittedIf permittedIf permittedIf permitted
Manage organisation usersNoNoYesYes
Manage prompts, schemas and templatesNoNoConfigured scopeYes
View audit and job monitoringOwn activityRelevant workOrganisation scopeAdministrative scope

AI governance

Assistance without invisible authority.

The model should show what documents were used, which workflow and version produced the output, what information is missing and who approved the final client-facing result.

Human judgement is not a hidden assumption.

It is a visible step with a named reviewer decision and a linked version.

  1. 1

    Eligibility check

    Confirm access, required information, documents, credits and that no conflicting analysis is running.

    Before start
  2. 2

    Visible progress

    Show queued, extracting, processing, analysing, validating, saving and report-generation stages.

    During work
  3. 3

    Structured result

    Present policy detail, exclusions, warranties, gaps, risks, recommendations, missing information and confidence context.

    Reviewable
  4. 4

    Reviewer decision

    Approve, approve with changes, reject or request more information with notes and quality concerns.

    Human control
  5. 5

    Version-locked report

    Generate the report from a selected analysis version and preserve every previous report version.

    Final proof

Security FAQ

Ask the harder questions early.

Production trust depends on verified implementation details, not interface copy.

Is Valentry certified to ISO 27001 or SOC 2?

No ISO 27001, SOC 2 or other third-party certification claim is made on this website. Independent assurance is stated only when formally verified.

Does Valentry support role-based access?

Role and permission management, organisation boundaries, assignment and administrative access are stated platform requirements. Exact permission sets are configured and tested during implementation.

What does immutable mean here?

Completed analysis versions should not be edited, final reports should not be overwritten and significant audit records should remain protected from alteration.

How are AI outputs controlled?

The workflow includes eligibility checks, visible progress, structured results, versioning, reviewer decisions and a rule that client-facing advice remains subject to authorised human review.

Where will data be hosted?

Client records and documents are stored on Google Cloud in the africa-south1 region (Johannesburg), so client personal information stays in South Africa. Subprocessors, backups, retention and incident processes are set out in the production agreement.

Is Valentry POPIA compliant?

The platform uses POPIA-aware handling and protection of personal information. Legal compliance also depends on implemented controls, contracts, operator relationships and each organisation’s own lawful processing practices.

Seven days free. R0 today.

Bring your compliance questions to the demo.

A serious product conversation covers permissions, audit events, versioning, AI review, report control, hosting and data-processing responsibilities.

Start your free trialBook a live demo
  • Purpose-built for insurance
  • Client data hosted in South Africa
  • Human-reviewed AI