No external LLM APIs
Company context stays inside the deployment. Diana does not call public model endpoints.
Diana Protect
Company-wide AI. Inside your control.
Give every person useful AI help while keeping company context, source permissions, and activity inside a controlled European deployment.
Controlled boundary
Give people useful AI. Keep company context inside.
Sources, context, AI help, and actions stay inside one deployment perimeter you define—dedicated EU cloud by default, with on-premise and air-gapped options when policy requires them.
Customer-defined control
Encryption keys, retention, access, and update procedures follow the deployment you select.
European by default
A dedicated sovereign EU cloud is the default path for regulated teams.
Identity-to-action control
One person. Context that fits them.
Diana evaluates the person, source, policy, and requested action as one continuous chain—so the AI help each person receives fits their access and role.
POLICY PATH · LIVE REQUESTENFORCED
REQUEST
Prepare a renewal risk memo from CRM and legal records
ACTOR
LEGAL COUNSEL · EU
01VERIFIED
IdentityResolve the actor
Workspace role and source identity establish who is asking.
02SCOPED
RetrievalIntersect permissions
Diana searches only the records that identity can already access.
03CITED
AnswerAttach the evidence
Permitted sources remain connected to every material claim.
04APPROVED
ActionApply the action gate
Policy decides whether work proceeds, waits for approval, or stops.
POLICY PREVIEWEffective access by actor and source
LIVE · REQUEST SCOPEDACTORCRMLEGALFINANCEEXPORT
Legal counsel ALLOW ALLOW MASK APPROVE
Finance lead ALLOW MASK ALLOW APPROVE
Research agent SCOPED SCOPED DENY DENY
Architecture · Access · Oversight
Control you can prove.
Company-wide AI needs control at every step—permissions, approvals, and visibility built into every search, answer, and action.
Each person gets the access they already have.
If someone cannot access a source outside Diana, they cannot retrieve it through Diana. Role-aware retrieval and scoped connectors keep existing access controls intact.
Make AI useful without creating shadow IT.
Approved sources, role-aware AI help, human approvals, controlled exports, and governed actions let people use Diana without losing oversight.
See how company AI is being used.
The Audit Stream records reads, edits, exports, shares, approvals, and automated activity so teams can inspect important activity.
Audit Stream
Evidence you can actually inspect.
Go beyond a policy statement. Inspect how information was accessed and how work was produced—reads, edits, exports, shares, approvals, and automated activity in one stream.
- Source and document lineage
- Human and agent activity
- Time-stamped approvals and exports
Deployment options
Choose the boundary your company requires.
Give every team the same Diana experience across dedicated EU cloud, on-premise, and air-gapped deployments—what changes is where the perimeter lives.
Dedicated EU cloud
A controlled European environment for regulated teams that need sovereignty without operating the full stack themselves.
- EU jurisdiction
- No external LLM APIs
- Customer keys available
On-premise
Run Diana inside your own infrastructure when hosting, network, or procurement rules require it.
- Customer infrastructure
- Scoped updates
- Local control plane
Air-gapped
Operate with no outbound connectivity when mission or classification requirements demand full isolation.
- No outbound network
- Offline update path
- Hard boundary
Operational control
Security that continues after go-live.
Company-wide AI is not a one-time deployment. Diana gives security and IT teams a clear operating model as more people and teams begin to use it.
Configure
POLICY BASELINEDefine the boundary
Select deployment, connectors, key custody, retention, roles, and approval paths before production access.
Observe
CONTINUOUS RECORDSee what the system does
Review access checks, cited outputs, exports, approvals, and automated activity through one evidence stream.
Change
GOVERNED CHANGEControl what evolves
Scope connector, model, policy, and software updates to the deployment process your organization approves.
Respond
END-TO-END TRACEInvestigate with context
Trace an event from actor to source, output, approval, and action without reconstructing evidence across tools.
Regulatory readiness
Built to produce evidence under review.
Diana helps teams show where processing happens, who could access what, and how company AI use is governed—without overstating certification claims.
GDPR architecture
Keep processing inside a controlled European deployment and produce evidence for access, retention, and purpose decisions.
DORA readiness
Give ICT and operational risk teams a concrete record of how AI systems access and transform critical information.
NIS2 defensibility
Support security and continuity reviews with permission-aware retrieval and a visible activity stream.
EU AI Act evidence
Document how outputs were produced, which sources were used, and which approvals governed the work.
EU HOSTEDNO EXTERNAL LLM APISPERMISSION-AWARESOURCE CITATIONSAUDIT STREAMCUSTOMER KEYSON-PREMISEAIR-GAPPED
Security review package
Give reviewers something concrete.
Move from a marketing conversation to a working review. We map Diana’s architecture and operating model to your questionnaire, policies, and deployment requirements.
Start a security reviewDIANA PROTECT · REVIEW WORKSPACECONFIDENTIAL
01SYSTEM REVIEW
Architecture walkthrough
Deployment boundary, trust zones, data paths, model placement, connectors, and administrative surfaces.
02FLOW REVIEW
Data-flow mapping
Where company context enters, how it is processed, what is retained, and where outputs are produced.
03CONTROL REVIEW
Control mapping
Permissions, approvals, audit evidence, key custody, retention, updates, and staff-access assumptions.
04FIT ASSESSMENT
Deployment assessment
A practical fit check across dedicated EU cloud, on-premise, and isolated operating requirements.
Available during a qualified security and architecture review.04 REVIEW MODULES
Technical proof
Answers security and procurement ask first.
Egress, training, permissions, keys, deployment, and documentation.
No. Diana runs without external LLM API calls, keeping company context inside the controlled European deployment.
Security briefing
Prepare company-wide AI for review.
Walk through architecture, access, retention, and audit with your security, compliance, and IT teams before Diana reaches more people across the business.