Dataceen & Architecture

Please check in on our FAQ section, there are some answers to common questions that might be of interest.

Dataceen application interface showing data modeling dashboard

Five ways to run it

The model is the same in all five. What changes is who runs the platform and where the data rests.

Hosted

Who runs the platform
Dataceen
Where the data rests
A database and a search index hosted by Dataceen, in a region we have open
Whose cloud account
Dataceen's
What you bring
Nothing

Hosted, your database

Who runs the platform
Dataceen
Where the data rests
A database and a search index you control
Whose cloud account
Dataceen's for the platform, yours for the data
What you bring
A supported database and an Elasticsearch deployment

Dedicated

Who runs the platform
Dataceen
Where the data rests
Database servers for your company alone
Whose cloud account
Dataceen's
What you bring
Nothing

Dedicated, in your cloud

Who runs the platform
Dataceen, inside your account
Where the data rests
In your own account
Whose cloud account
Yours
What you bring
A cloud account on Azure, AWS or Google Cloud

Self-run

Who runs the platform
You. We set it up with you
Where the data rests
Yours
Whose cloud account
Yours
What you bring
Your own accounts for the outside services it uses, such as AI and maps

Regions. Hosted and Hosted, your database run in the regions we have open, the EU by default. Dedicated, Dedicated in your cloud and Self-run can be in any region on Azure, AWS or Google Cloud.

The search index goes with the database. When you control the database, you control the search index too. No copy of your records rests with Dataceen.

Reading your data directly. With Dedicated and beyond, your reporting and analysis tools can read the database directly. Changes should go through the API: that is where the model's rules, the permissions and the call log apply. Something written straight into the database bypasses all three.

Self-run. You get development, test and production environments of your own. They use your own accounts for AI and for maps, and they keep running without a connection to Dataceen. Self-run is licensed per year.

Learn more about how Dataceen works

The Business Data Model

The Business Data Model is the core of Dataceen. It defines how business data is structured and connected, independent of where the data physically resides. It is created and managed using the Dataceen Administration Tool.

A model describes what the business data is and how it relates, without considering where that data physically resides. The actual data may exist in many different systems — such as ERP, CRM, vehicle registries, customer databases, or other master data sources — but the model provides a unified business view. Dataceen can contain any number of models, each varying in size and complexity.

A Graph-Based Structure

Each model is structured as a graph, consisting of:

  • Nodes — representing business entities such as Customer, Vehicle, Order, Technician, or Service Appointment
  • Relationships — describing how entities are connected (e.g., Customer owns Vehicle or Technician performs Service)
  • Properties — attributes that describe nodes or relationships, such as name, registration number, service date, or status. Properties can be sourced from any master data system, allowing a single node to combine data from multiple systems.

This graph structure creates a clear and intuitive representation of the business domain, making it easier for both humans and systems — including AI — to understand and work with the data.

Flexible Model Creation: AI, Manual, or Combined

  • AI-assisted creation — AI can analyze descriptions, documents, or existing data structures and generate an initial model automatically.
  • Manual design — Architects and domain experts can build or refine models manually using the administration tool, ensuring full control over structure and meaning.
  • Hybrid approach — AI generates the initial structure while experts refine, extend, and validate the model. This enables rapid development while maintaining precise business control.

Regardless of how the model is created, the result is always a fully controlled and transparent data structure owned by the enterprise.

A Contract for Data Integration

The Business Data Model acts as a contract between systems. It defines how data from different sources is integrated, interpreted, and used across applications and services. Because the model knows exactly which system owns each data element, Dataceen always knows where to retrieve the correct data.

Automatic Platform Generation

Once a model is created, Dataceen generates from it:

Secure APIs
Event streams
Search indexes
Data access services
AI-ready MCP servers
Comprehensive documentation

Security and access control are also derived directly from the model. This means the Business Data Model does not just describe the data — it drives the entire Dataceen platform, ensuring that all systems access and use data in a consistent, secure, and well-defined way.

Challenges architects face every day

Data integration shouldn't be a full-time job. Yet architects spend most of their time solving the same structural problems over and over.

Fragmented data landscape

Business data is spread across dozens of systems with no unified model. Every integration project starts from scratch.

Months-long integration projects

Each new data need requires custom APIs, security layers, and documentation — all built manually, every time.

Inconsistent security & governance

Access control is implemented differently across systems. There's no single source of truth for who can access what.

How Dataceen fits into your architecture

Dataceen sits between the systems you already run and the applications, agents and partners that consume data. Each existing system is met on its own terms: replaced, fronted, fed in, fed out, or kept behind an adapter.

ERP

CRM

BI Tools

Custom Systems

Dataceen

Shared data model · Auto-generated infrastructure

APIs

Security

Search

Documentation

Applications

Dashboards & BI

Partner integrations

Applications are built by your team on the SDK, or built and hosted by Dataceen: a dashboard, a site or a task program, tested by a gate and live in 10 to 40 minutes, code in a GitHub repository you can take over.

Five ways to meet an existing system

Decided per system, not for the whole landscape.

Replace it

Its data moves into a model; its users move to a new application.

Front it

A new application or API in front, the old system behind.

Feed in

The old system keeps running and feeds the model.

Feed out

The model holds the data; the systems that stay follow it through events or a SQL mirror.

Keep the interface

An adapter serves the old API or file format one to one, so partners change nothing.

Most first projects start with feed in plus one new application, and replace later: low risk, value early.

Designed for real-world architectures

Here's what that means in practice.

Replace, front, feed in, feed out or keep the interface: decided per system, so the first project carries little risk.

No vendor lock-in. Use your own databases, your own cloud, your own infrastructure.

A model is built in three to twenty minutes from a description. An application is live in 10 to 40 minutes.

Raw access isn't enough

Every enterprise AI project hits the same wall. The agent can reach the data, but three systems disagree on what a customer is, nobody has written down which fields it may change, and the permissions live in integration code and people's heads. That isn't a data quality problem. It's missing context.

With raw data access

  • The agent guesses what fields mean
  • Definitions differ between systems
  • Permissions are reimplemented for every consumer
  • Every new app or agent is a new integration project
  • No record of what the agent did or why

With Dataceen

  • Entities, relationships and rules are defined once
  • The same meaning in the API, the SDK, the search index and the MCP tool
  • Permissions come from your directory and apply to humans and agents alike
  • New consumers reuse the model instead of rebuilding it
  • Every change is recorded as an event you can replay and audit

Security and control

Your business controls where Dataceen-managed data is stored.

Customer-controlled storageRole-based accessFine-grained permissionsAudit logsPrivate by designCompliance-friendly architecture

Sound familiar?

Four situations we hear every week. Each one ends the same way: describe the data once, and Dataceen generates what the situation needs.

“Management wants AI on our data. Security keeps saying no.”

Both are right. Dataceen exposes your data through MCP tools that inherit your directory's roles and permissions — every call logged and auditable. AI agents get access the way people do: safely.

Start from: a database schema or API spec.

“Answering our customers' questions takes three systems and two phone calls.”

Describe the data once. Dataceen generates the API, the search and the AI tools that answer plain-language questions from live data — under the permissions that already exist.

Start from: API specs, schemas or documents.

“The portal has been postponed twice. Every integration is quoted in months.”

From your model, Dataceen generates the system — APIs, typed SDKs, documentation and ready-made instructions for coding agents. The system you've postponed becomes days of work, not quarters.

Start from: what your systems already expose.

“Honestly? We're tired of AI pitches. We need integrations, permissions and audit sorted.”

Fair. Dataceen doesn't require an AI strategy: generated APIs, fine-grained access control and a complete event trail solve the everyday problem. And if you ever do want AI on your data — it's already safe.

Start from: a database schema, or even a PDF describing the domain.

Ready to simplify your data architecture?

Build a first model from a description, and decide per existing system how it fits in.