SentraMind
中文 Private Evaluation

Sovereign AI security · Southeast Asia

Stop sensitive data reaching public AI without sending security data to another cloud.

SentraMind gives security teams a customer-controlled path from signal to deterministic policy decision and inspectable evidence. Platform Core and GenAI Prompt DLP are available for a private proof of concept.

Request a Private EvaluationRun the Live Demonstrations
Cutaway of a customer-controlled environment where the security decision and evidence remain inside the boundary and a prohibited external route terminates at its edge.
Customer-controlled environment
  1. Prompt submitted
  2. Deterministic detection
  3. Customer policy
  4. External route stopped
  5. Illustrative decision receipt
Prompt warned POC example
Authority
Customer policy
Provenance
Prompt submitted
Event time
Policy reference
PROMPT-DLP / POC-04
Synthetic example Integrity not asserted

The operating problem

Public AI creates a new path out of your control boundary.

Employees move quickly. Sensitive prompts can include credentials, personal data, source code and operational context before a conventional review process can react. For sovereign and regulated environments, adding another hosted security layer can recreate the same disclosure problem.

  1. 01 Keep the security decision inside infrastructure you control.
  2. 02 Make deterministic policy authoritative; model assistance remains subordinate.
  3. 03 Give reviewers a readable record of what happened and why.

What you can evaluate now

A bounded POC, built around one consequential workflow.

The evaluation is deliberately narrow: prove the boundary, policy path and evidence before expanding scope.

Available for POC

Platform Core

The governed runtime, policy path and evidence foundation used by the POC.

Available for POC

GenAI Prompt DLP · Sentry

Deterministic prompt inspection and customer-defined warn or block outcomes.

Coming next

AI Usage Observability · Cartographer

Visibility into organisational AI usage is the next planned capability, not part of today’s POC.

From event to evidence

The control path stays understandable under scrutiny.

  1. 01

    Observe

    A prompt reaches the controlled inspection point.

  2. 02

    Detect

    Deterministic detectors identify policy-relevant content.

  3. 03

    Decide

    Customer policy authorises the outcome; assistance cannot overrule it.

  4. 04

    Record

    A decision receipt preserves outcome, authority and provenance.

Deterministic browser demonstration

Test the detector layer before the evaluation.

The live demonstrations execute deterministic detector logic locally in your browser. They do not reproduce the appliance, model classification or a production deployment—and they are not evidence of those systems.

Open all live demonstrations

Qualified trust evidence

Proof should identify its boundary.

01

Customer-controlled deployment

Designed for self-hosted and air-gap-capable operation. Final topology is agreed and tested during evaluation.

02

Authoritative policy path

Deterministic controls and customer policy remain authoritative; model assistance is not permitted to make consequential decisions.

03

Inspectable records

Current code includes hash-chain and evidence-sealing primitives. A complete end-to-end evidence pipeline remains evaluation work.

04

First-party portal runtime

This portal is built without hosted fonts, analytics or third-party runtime assets. Deployment network behaviour is verified separately.

The signature artifact

Every governed outcome should end in a receipt a reviewer can interrogate.

A decision receipt makes the policy authority, source, event time and integrity state explicit. The example shown here is synthetic and demonstrates the intended evidence grammar—not a certificate and not a production event.

Prompt warned POC example
Authority
Customer policy
Provenance
Prompt submitted
Event time
Policy reference
PROMPT-DLP / POC-04
Synthetic example Integrity not asserted

Capability horizon

A focused product now; a governed crew over time.

Cartographer is coming next. The remaining specialist concepts—from email and source-code review to vulnerability and incident workflows—belong to the M3–M9 design-partner roadmap. They are published to guide discovery, not represented as available features.

Explore the Crew roadmap →
Available for POC Platform Core + Sentry Coming next Cartographer Design-partner roadmap Crew M3–M9

Built for regional operating reality

Start with sovereignty requirements, not a generic cloud assumption.

The evaluation conversation covers site constraints, disconnected operation, ownership of policies and evidence, and the regulatory context relevant to Southeast Asian public-sector and regulated buyers. Regulatory material supports discovery; it is not legal advice.

Review regional context

A private, evidence-led evaluation

Bring the boundary, workflow and acceptance criteria that matter to your team.

We will scope a POC around GenAI Prompt DLP, identify what your environment must provide, agree measurable acceptance criteria and document what is current, illustrative or still to be built. Your security, architecture and procurement stakeholders leave with one shared record of the evaluation boundary.

Request a Private EvaluationReview deployment options