
How IDZ Stands Out
Most AI assistants are cloud-first and hard to audit. MetaCortex works across your encrypted files and chats with on-device processing where supported, Boost with consent, and receipts for every result.

Processing mode
On-device where supported; Boost with consent
Typically cloud-first by default
Users see and control when data leaves the device
Working on your private data
Searches all encrypted files, folders, and chats. No manual uploads.
Often require uploading files onebyone or pasting text per chat; connectors vary by setup and permissions
Faster answers, fewer risky uploads, fewer permissions to manage
Evidence & citations
Answers cite exact sources (file/page and chat message)
Evidence trails vary and aren’t always sourceanchored
Faster verification, fewer hallucinations
End-to-end AI workflows
Find (cited retrieval) → Act (transform/redact/package) → Share with controls → Proof via signed receipts
Often stop at “answers”; completing the workflow requires manual handoffs to other tools
Less busywork, fewer mistakes, and safer sharing endtoend
Saved artifacts (shareready outputs)
Each action produces a saved artifact (report/redacted file/CSV) you can reopen and share later
Outputs often live in chat; saving, versioning, and sharing are manual and inconsistent.
Reusable work products, easier collaboration, and fewer “lost” AI results.
Receipts & auditability
Signed receipt per run (mode, sources, actions, policies). Exportable.
Audit trails vary; often limited to chat history/admin logs.
Makes AI usage provable for compliance and review.
External data processing
Admin policies for Boost, memory retention, and sharing.
Controls exist but vary by platform and integration.
Consistent governance across the organisation.
Customer-hosted AI endpoint (Enterprise)
Route Boost to your on-prem/VPC model endpoint (no public cloud).
Usually tied to provider infrastructure or limited deployment options.
Enables regulated teams to adopt AI safely.
Memory (personalization)
Memory is configurable (Never / Timelimited / Always) and policycontrolled in Enterprise.
Memory controls vary by product and plan.
Predictable retention and safer personalization.
Processing mode

On-device where supported; Boost with consent
Typically cloud-first by default
Users see and control when data leaves the device
Working on your private data

Searches all encrypted files, folders, and chats. No manual uploads.
Often require uploading files onebyone or pasting text per chat; connectors vary by setup and permissions
Faster answers, fewer risky uploads, fewer permissions to manage
Evidence & citations

Answers cite exact sources (file/page and chat message)
Evidence trails vary and aren’t always sourceanchored
Faster verification, fewer hallucinations
End-to-end AI workflows

Find (cited retrieval) → Act (transform/redact/package) → Share with controls → Proof via signed receipts
Often stop at “answers”; completing the workflow requires manual handoffs to other tools
Less busywork, fewer mistakes, and safer sharing endtoend
Saved artifacts (shareready outputs)

Each action produces a saved artifact (report/redacted file/CSV) you can reopen and share later
Outputs often live in chat; saving, versioning, and sharing are manual and inconsistent.
Reusable work products, easier collaboration, and fewer “lost” AI results.
Receipts & auditability

Signed receipt per run (mode, sources, actions, policies). Exportable.
Audit trails vary; often limited to chat history/admin logs.
Makes AI usage provable for compliance and review.
External data processing

Admin policies for Boost, memory retention, and sharing.
Controls exist but vary by platform and integration.
Consistent governance across the organisation.
Customer-hosted AI endpoint (Enterprise)

Route Boost to your on-prem/VPC model endpoint (no public cloud).
Usually tied to provider infrastructure or limited deployment options.
Enables regulated teams to adopt AI safely.
Memory (personalization)

Memory is configurable (Never / Timelimited / Always) and policycontrolled in Enterprise.
Memory controls vary by product and plan.
Predictable retention and safer personalization.
Key takeaways
True Privacy
IDZ’s zero-knowledge approach and user-held encryption keys ensure that nobody, not even IDZ, can access your data.
Easy Yet Secure
Passwordless authentication, no personal data requirements, and integrated features make privacy straightforward for everyone.
Self-Managed
By putting you fully in charge of key management and data recovery, IDZ removes all external parties from the equation.