Where does it run?
Inside the environment you control: your cloud subscription, existing analytical platform or on-premise data centre.
For your technical team
Your platform, security and data engineering teams will operate this. Here is what runs where, what it needs from your estate, what it changes in your team's work, and what never changes: business data stays inside your perimeter, and nothing executes without your approval.
Data Masters operates inside the client-controlled environment, above your existing data assets and below governed outputs. It ships as a VM image and runs inside your perimeter; business data stays inside, the AI layer sees metadata only and can be switched off, and credentials, policies and deployment approval remain with your team.
Architecture at a glance
The deployment model is designed to make the security conversation concrete before procurement begins.
Inside the environment you control: your cloud subscription, existing analytical platform or on-premise data centre.
Your records, tables and analytical assets remain inside your boundary. Data Masters Decision Intelligence works where those assets already live.
Access is scoped to the selected sources, workspaces and delivery targets. Credentials and permission policies remain customer-controlled.
Proposed changes are versioned, impact-checked and presented for approval before controlled release. The evidence stays attached.
The deterministic core continues to model, build, govern and activate. AI accelerates metadata interpretation and suggestions; it does not execute decisions and is not required to run the core.
Before the evaluation
You can, and the question is capacity. Count the requests waiting behind your team today, and measure the time from a business request to something in production. Decision Intelligence runs the pipelines, models and checks from declared definitions, so first results land in the first month, and your engineers spend their days on the work only they can do.
Decision Intelligence ships as a VM image into your environment and runs inside your perimeter, on your identity and network controls. The core is deterministic and runs with the AI switched off, and the AI layer sees metadata only. Nothing executes without a human approval: every change lands as a proposal in an inbox, versioned and impact-checked, and your team releases it.
One standard way of building, for everyone. Definitions, pipelines, models and checks come from the same configuration, so there is no personal coding style to inherit, and a new engineer is productive in the first weeks. Training is part of the delivery and has an end date, so the standard is yours to keep.
It replaces work. The pipelines, the catalog, the quality checks and the semantic model are maintained in one place, from one definition, and a change to the definition flows through everything downstream. What you retire is the hand-written glue between four tools.
Controlled delivery
Business intent and technical evidence travel together through proposal, impact check, approval and release.
Produce the model, mapping or action as a visible proposal.
Show dependencies, affected assets and the reason for change.
Route every change to the accountable person for approval.
Deploy a versioned package through the environment you operate.
Keep the proposal, approval, lineage and outcome connected.
Deployment choices
The target architecture is agreed with your platform and security teams before technical delivery begins.
Add the decision layer above an established warehouse, or a Microsoft Fabric, Databricks, Snowflake, AWS, or MinIO with on-premise Spark environment.
Run within the cloud boundary, identity model, network controls and operating processes your team already manages.
Operate inside your data centre when policy or regulation requires an entirely customer-hosted deployment.
We will provide the material your teams need to assess Decision Intelligence without filling the page with assurance badges.
Request the security and architecture packTechnical evaluation
We map how Decision Intelligence reads it, what it builds, and what your team approves, on your infrastructure.

We use cookies and similar technologies to see how the site is used, so we can improve it. Choose "Accept all", or open preferences to pick which categories you allow. Necessary cookies are always on; anything else runs only with your consent, and you can change or withdraw it at any time.