SentraMind

Sovereign AI security · Southeast Asia

Protect sensitive data before it reaches public AI.

Explore prompt inspection and customer-defined policies inside infrastructure you control.

Platform Core + Sentry Available for POC
Try browser demo

Browser demos explore deterministic detector logic.

Illustrative POC example
Synthetic data
Sentry GenAI prompt inspection
01 / 03
SAMPLE PROMPT

Summarise [customer records]
and prepare a quarterly report.

Customer data pattern
POLICY OUTCOME
Block

Example policy stops this prompt.

DECISION RECORD
Authority
Customer policy
Outcome
Block
Source
Synthetic example
Interactive illustration · not a connected deployment
Your policy is the authority. SENTRAMIND / SENTRY

Inspect sensitive prompts

Start with policy-relevant patterns.

Apply your policies

Keep the decision under your control.

Review decision records

Understand the outcome and its authority.

What you can evaluate

Start with a focused evaluation.

A clear starting point. A defined scope.
A practical way to establish what works for you.

Available for POC

01 / FOUNDATION

Platform Core

The foundation for customer policies, controlled decisions and reviewable outcomes.

Policy controlsDecision records
Coming next

03 / CARTOGRAPHER

AI Usage Observability

A planned view of organisational AI usage to help identify where controls are needed.

AI usage visibilityPlanned capability

Current capabilities and future plans are clearly separated. Explore the scope before planning a deployment.

From input to outcome

A policy decision you can follow.

A straightforward control path, with customer policy at its centre.

01

Observe

A prompt reaches the agreed inspection point.

02

Detect

Deterministic checks identify policy-relevant patterns.

03

Decide

Your policy determines the response. Assistance cannot overrule it.

04

Record

Review the outcome, policy reference and source in a decision record.

Start with the work

Everyday AI use.
Consequential decisions.

Bring a real concern. Use synthetic or approved examples to define a clear test.

01

Customer information

Assess how defined customer-data patterns should be handled before a prompt reaches public AI.

SAMPLE INPUT Summarise [customer records]
02

Credentials in copied text

Explore policy responses to sample access keys, tokens and explicitly defined sensitive patterns.

SAMPLE INPUT Review [example access key]
03

Appropriate AI use

Compare a flagged example with an ordinary request to understand how policy outcomes differ.

SAMPLE INPUT Draft a public announcement.

These synthetic scenarios illustrate the workflow. Coverage and expected outcomes are agreed during the POC.

Your environment. Your requirements.

Security decisions belong
inside your boundary.

Define where processing happens, who controls policy and how evidence is handled. Start with the environment you need to protect.

  • Customer-controlled infrastructure
  • Policy ownership and access rules
  • Defined evidence and retention requirements
Explore deployment considerations
CUSTOMER-CONTROLLED ENVIRONMENT
Your infrastructure
SentraMind Controlled inspection & policy
Your policies
Your evidence

Deployment concept · final topology agreed in the POC

A clear starting point

A few practical
questions.

Know what you are evaluating before you begin.

What can we evaluate today?

Platform Core and Sentry are available for a scoped POC. Cartographer is coming next. Other specialist capabilities remain on the design-partner roadmap.

Is the interactive example the complete product?

The example on this page uses preset synthetic scenarios to explain the workflow. The linked browser demonstrations exercise deterministic detector logic. Neither represents a complete production deployment or model-classification system.

Is private model inference available now?

Private AI Core describes the intended architecture and sizing approach. The current portal describes a build with no configured model runtime; this capability remains on the design-partner roadmap.

How do we start an evaluation?

Start by reviewing the POC scope and deployment requirements, then explore the browser demonstrations. Evaluation intake is not yet enabled on the current portal.

Clarity before commitment

See the decision.
Understand the control.

Explore the demonstrations and what a focused POC can establish.

Try browser demo
Available for POC

A foundation for governed decisions.

Platform Core provides the policy and evidence foundations for the current POC. Define who controls the rules, follow how an outcome is reached and agree what reviewers need to see.

  • Customer-defined policy and decision paths
  • Inspectable outcomes and source references
  • A bounded evaluation with agreed acceptance criteria
Read the technical overview
Available for POC

Put prompt policies to the test.

Use an agreed set of synthetic or internally approved examples to evaluate deterministic prompt inspection and customer-defined warn or block outcomes.

  • Define sensitive patterns and expected outcomes
  • Review the policy authority behind each decision
  • Document coverage, false positives and known exclusions
Open the prompt sandbox
Coming next

A clearer view of organisational AI use.

Cartographer is the next planned capability: visibility into where and how AI is used, to inform decisions about controls and priorities.

  • Planned organisational AI usage visibility
  • A basis for discussing coverage and control gaps
  • Defined separately from today's Core and Sentry POC
Read the product roadmap
Platform Core + Sentry

A focused POC. A shared definition of success.

Start with one workflow, your operating constraints and clear acceptance criteria. The evaluation establishes what the agreed scope can do in your environment.

  1. Define the boundary Agree the environment, owners, data classes and expected decisions.
  2. Test approved examples Review detections, outcomes, false positives and limitations.
  3. Review the evidence Document results and decide the appropriate next step.
Read the evaluation guide