See what leaves.
Control what leaves.
Prove what happened.
Before external processing, review and edit the sanitised data. After every result, open a verified receipt showing the route, sources and actions.
AI should not operate behind an invisible boundary.
When AI works across sensitive files and chats, you should know what was shared, which sources were used and how the result was produced. MetaCortex makes those steps visible.
What information left the workspace?
Which sources were used?
Which processing route handled the request?
What actions were performed?
Review before anything is sent
MetaCortex shows the sanitised prompt and source excerpts before Boost or Company AI processing. Remove a complete source, paragraph or selected sentences, then approve or cancel.
- Review the exact sanitised text
- Remove entire sources or selected passages
- Approve once or for the thread
- Cancel without sending anything
Every processing route is clearly identified
- 01
Encrypted workspace
- 02
User-approved sanitised context
- 03
Approved AI route
Private
Runs locally on supported devices. No request content is sent to external AI.
Boost
Approved sanitised context is sent to IDZ-managed cloud AI.
Company AI
Approved sanitised context is sent to the organisation’s configured AI endpoint.
Web-assisted
Live web search is clearly labelled and recorded.
A verified receipt for every result
The receipt records how the result was produced and can be opened directly from the response.
Receipt
Boost • 2 sources • No web search • 3 actions
Privacy & compute
Boost • Balanced • IDZ-managed cloud AI Data sent – Sanitised prompt, 2 files
Input & Sources accessed
Sources – 2 files • No web search Removed by you – 2 sources, 2 sentences
Actions performed
Summarized (5 bullets) Redacted (6 items) Compared (v2 vs v3)
Output & sharing
Saved to ZCloud
Proof
Proof – #548595 • Jan 7 2026, 11:26:32
- 1Mode, model and provider
- 2What was sent and how it was approved
- 3Internal sources and web use
- 4What you removed before sending
- 5Actions performed
- 6Where the result was saved or shared
- 7Receipt ID, signature and verification status
Why it matters
For organisations
- Apply policy across members and AI routes
- Make AI-assisted work reviewable
- Support audits, investigations and usage checks
For individuals
- Know what leaves before approving
- See the sources and processing route
- Keep a record with the work
Receipt use cases
Institutional privacy. A boundary you can see.
IDZ combines institutional-grade safeguards with a privacy boundary users can see, control and verify.
End-to-end encryption & customer-managed keys
Files and chats are end-to-end encrypted, and customers retain control of the keys needed to access them.
Regional processing & tenant isolation
Regional storage and processing options, isolated workspaces and restricted internal access reduce unnecessary exposure.
Retention, deletion & no training
Customers control retention and deletion. Approved AI processing is contractually excluded from model training, with zero-retention configurations where supported.
Enterprise governance & auditability
Roles, policies, monitoring, audit logs and verified receipts make AI use controllable and reviewable.
Governed subprocessors
Subprocessors are selected, contracted and reviewed under IDZ security and data-protection requirements.
Independent assurance
Independent testing, including Cure53, and progress toward ISO 27001 and SOC 2 are available in the IDZ Trust Center.
Ready to work with private AI on your real data?
Download IDZ and meet MetaCortex.