Teh Tarik Digital · Solutions · Healthcare & Clinical Compliance

The Compliance Question Healthcare AI Has to Answer First

Health data is sensitive personal data under PDPA Section 40, and since the 2024 Amendment, that obligation extends directly to processors, not just the healthcare provider. That makes compliance the starting constraint for healthcare AI, not a feature added later — getting it wrong isn't a UX problem, it's a regulatory one.

Teh Tarik Digital's healthcare AI work is built the other way around: architecture designed to the compliance requirement first, features second. This is the same approach behind NURI, our clinical AI platform — the depth of that work is what this page draws on.

2.5+ hrs
admin time reclaimed per clinician shift
40%
reduction in documentation turnaround
100%
sovereign architecture, zero external exposure
Q3 2027
pilot deployment target

Who This Is For

// FIT

Built for operators who need a straight answer on compliance before they'll take a second meeting — not a generic clinical-AI pitch. If a vendor can't tell you who's the Data Controller and who's the Processor before you've signed anything, that's not a detail to sort out later; it's the whole risk profile of the engagement.

Private clinics and hospitals evaluating AI-assisted documentation or triage.
Residential aged care operators managing shift handovers and incident escalation.
Any healthcare operator that needs to know, in writing, who's the Data Controller and who's the Data Processor before adopting an AI tool.

Core Capabilities

// CAPABILITIES

Four capabilities, each one designed around the compliance obligation first — not retrofitted after the fact. That ordering isn't a nice-to-have in healthcare; it's the difference between a tool a compliance officer can actually sign off on and one that gets stuck in review for months while someone tries to reverse-engineer where the data actually goes.

01

Voice Clinical Documentation

Dictated notes structured into records automatically — built to retrieve and structure from your actual dictated notes, not locked to a single language or a fixed template.

02

Early-Warning Scoring

Automated deterioration scoring (NEWS2-aligned) routed to the duty officer already defined in your escalation chain — the system flags and routes, it doesn't override clinical judgment.

03

PDPA-Aligned Architecture

Data Controller/Processor roles defined up front, not left for a Terms of Service page to sort out later; on-premise or Malaysian-region deployment; zero external data dependency by default.

04

MOH-Aligned Workflows

Built around local documentation standards and clinical escalation patterns from the start — not a Western scribe tool with Malaysian terminology bolted on afterward.

The compliance question most vendors don't answer: before adopting any clinical AI tool, an operator needs a straight answer to who is the Data Controller, who is the Processor, what's the legal basis for processing sensitive personal data under s.40, and what happens on a breach under s.12B. Our NURI Privacy Notice answers all four in writing — worth reading even if you're evaluating a different vendor, as a benchmark for what a healthcare AI vendor should be able to tell you.

Benefits & Results

// PROOF

NURI is pre-pilot — we're not going to overstate where it stands. What's actually built (the sovereign LLM, voice engine, EWS analytics) and what's still ahead (HIS integration, the Q3 2027 pilot) are two different lists, and we're keeping them separate rather than blending them into one confident-sounding pitch.

What Changes

  • Reduced manual charting timeVoice-driven documentation structures notes automatically.
  • Faster escalationAutomated NEWS2-aligned deterioration scoring routes to the right duty officer without delay.
  • Compliance defined up frontData Controller/Processor roles documented before deployment, not left ambiguous.
  • No rip-and-replaceIntegrates with existing clinical software.

Where NURI Stands

  • 2.5+ hrs / 40%2.5+ hours of admin time reclaimed per clinician shift, and a 40% reduction in total documentation turnaround.
  • Zero external exposure100% sovereign architecture.
  • Built vs. in progressCore sovereign LLM, voice engine, and EWS/incident analytics are built; HIS/facility system integration is in progress. Pilot deployment is targeted for Q3 2027, MOH/compliance reporting following in Q4 2027.

How It Works

// PROCESS

Compliance gets architected before a single feature ships — not the other way around. A clinical AI tool built feature-first and patched for compliance later is a liability wearing a product's clothing; we start from the s.40 obligation and build the feature set inside that boundary.

01

Assess

Clinical workflow audit, data flow mapping, Controller/Processor role definition — this is the phase most vendors skip past to get to the demo, and it's the one that actually determines whether compliance holds up later.

02

Architect

On-premise or Malaysian-cloud deployment plan, PDPA s.40 legal-basis documentation — written and signed off before deployment, not assumed and revisited after a breach.

03

Deploy

Voice documentation, escalation scoring, handover automation live on your systems — integrated with what you already run, not a separate tool your clinicians have to context-switch into.

04

Operate

Ongoing support under a defined processing agreement, not an open-ended SaaS relationship — the Controller/Processor split from stage one stays enforceable for the life of the engagement.

// FROM THE LAB

Real Builds, Real Numbers

Not healthcare-specific, but the closest proof we have — compliance-grade document extraction under real regulatory pressure.

DOCUMENT PROCESSING · COMPLIANCE

91% Accuracy, Confidence-Scored for Human Review

How we built a compliance-aware extraction pipeline with per-field confidence flagging — the same discipline sensitive healthcare data requires. Full case study →

FAQ

// 9 QUESTIONS

The compliance questions that come first, every time — answered honestly, including where NURI is still pre-pilot. We'd rather you read a "not yet" here than find out the hard way after signing.

Who is the Data Controller vs Processor?
For most clinical deployments, the healthcare facility is the Data Controller and Teh Tarik Digital acts as the Data Processor — this split is defined in writing before deployment, not left ambiguous.
Is this compliant with PDPA s.40?
Health data is sensitive personal data under PDPA s.40. Our architecture is designed around that requirement from the start — legal basis for processing is documented as part of deployment, not assumed. See our broader PDPA data-privacy breakdown.
Do you replace our existing clinical software?
No — voice documentation, escalation scoring, and handover automation integrate with what you already run.
What happens on a data breach?
Breach notification follows PDPA s.12B requirements; this is documented as part of the processing agreement before deployment, not decided after the fact.
Is NURI already deployed in Malaysian hospitals?
Not yet — we're not claiming a broader deployment track record than that. Core architecture (sovereign LLM, voice engine, EWS/incident analytics) is built; HIS/facility integration is in progress, with pilot deployment targeted for Q3 2027.
How does the escalation scoring decide who gets notified?
Deterioration scoring routes to the duty officer defined in your existing escalation chain — the system flags and routes, it doesn't override clinical judgment. See our human-in-the-loop design breakdown.
Does voice documentation work with Malay-language clinical notes, or English only?
The documentation pipeline isn't language-locked — it's built to retrieve and structure from your actual dictated notes rather than assume a single language. See our RAG pipeline breakdown.
What's the difference between this and a generic medical scribe or dictation tool?
A generic scribe transcribes; NURI's architecture is built around the compliance obligation first — Controller/Processor roles and legal basis are defined before a single note is dictated.
Can this extend to non-clinical healthcare operations — billing, scheduling, compliance logging?
The same document-intelligence and workflow approach applies to healthcare-adjacent operational workflows, though clinical use is where the compliance bar is highest. See our tender and compliance document processing project log.

Tell us your facility type and workflow — we'll show you exactly how the compliance split works.