Capabilities
What we build with, across every venture.
Every venture in the foundry draws on the same five disciplines. This is the instrumentation behind the ventures, not a menu of services we sell separately.
01 — Discipline
Applied AI Systems
We build production LLM pipelines, not demos. A pipeline that runs a law firm's transcription queue or a survey firm's document intake has to hold up against volume, edge cases, and the cost of a wrong output, day after day.
Model orchestration is domain-specific: the right model for the step, the right prompt for the document type, and a hybrid AI/human QA layer that catches what the model should not be trusted to decide alone. We treat AI as one stage in a production process, not the whole of it.
Instruments — Production LLM pipelines · Hybrid AI/human QA · Domain-specific model orchestration · Proprietary editors (VTTE, TRE)
02 — Discipline
Workflow Engineering
Most software describes a process. We enforce one. A workflow we build models the phases a job must move through explicitly, and the transitions between them are gates: a step that skips its prerequisites is blocked, not flagged.
This is structural, not a feature toggle. One of our production designs governs a regulated job across eleven phases and twelve hard gates, with an audit trail at every transition — so a firm's obligations are held by the system, not by whoever happens to be watching.
Instruments — Enforced job flows · Hard gates between phases · Audit trails per transition · Phase-gated process modelling
03 — Discipline
Automation
Once a workflow is governed, the routine work inside it can be automated safely. We build document pipelines and transcription automation that know the state of a job and what each phase requires, so the system assembles what used to be assembled by hand.
The same discipline extends to integration: our pipelines connect into government and enterprise systems of record, so automation does not stop at the edge of one application.
Instruments — Document pipelines · Transcription automation · Government system integration · Enterprise system integration
04 — Discipline
Enterprise Infrastructure
We engineer the infrastructure a vertical product runs on, not just the product itself. That includes telephony — PABX and IP voice systems — for enterprise and government deployments where communications infrastructure is part of the operating requirement.
We run self-hosted and cloud architecture depending on what a client's environment and data-residency posture demand, and we hold deployment and handover to the same standard: documented, reproducible, and owned by the client at the end of it, not by us.
Instruments — Telephony (PABX / IP) · Self-hosted architecture · Cloud architecture · Deployment & handover discipline
05 — Discipline
Compliance & Assurance
We are running an ISO/IEC 27001 information security programme and an ISO/IEC 42001 AI management programme, both in progress. We state that plainly: in progress, not certified, and we will update this line the day that changes and not before.
Malaysian government clients bring data-residency requirements that are not negotiable, and we design for them from the first architecture decision rather than retrofitting compliance once a system is already deployed.
Data storage follows a sovereign principle. Every application we operate runs in the AWS Malaysia Region, so client data remains in-country by default rather than by exception. The infrastructure underneath is enterprise-grade, engineered for stability under the load these clients place on it.
Instruments — ISO/IEC 27001 — information security (in progress) · ISO/IEC 42001 — AI management (in progress) · Data-residency awareness, Malaysian government clients · AWS Malaysia Region (data residency)
Building in a vertical
See the model behind these disciplines.
These five disciplines are how we operate every venture, from identification through certification. The full method is set out on the foundry page.