Residency
The physical location of the data and of the systems that process it. It is the easiest condition to establish and the one every infrastructure provider offers through geographic regions.
Where the data resides, whose jurisdiction reaches whoever holds it, and on what terms it can be moved to another system. This page describes how each of those three questions is resolved at Opendome.
«Data sovereignty» is frequently used as a synonym for storage in Europe. Strictly speaking it describes three distinct conditions, which can be met separately. An organisation assessing a provider needs an answer to all three.
The physical location of the data and of the systems that process it. It is the easiest condition to establish and the one every infrastructure provider offers through geographic regions.
The legal system that binds the entity controlling the data. It depends on the provider’s nationality and ownership structure, and is not determined by where the servers sit.
The ability to move the data to another system and carry on operating. It depends on the storage format and on the contractual terms for extraction and termination of the service.
The distinction matters in practice: a European region establishes the first condition and says nothing about the other two. A good share of the discrepancies between a project’s commercial assessment and its risk assessment originate in that difference.
The Clarifying Lawful Overseas Use of Data Act (CLOUD Act), in force in the United States since 2018, empowers its authorities to compel data from entities subject to its jurisdiction regardless of the country where it is stored. The location of the data centre does not change that reach.
What determines whether a company must answer a US demand is where it is incorporated and who controls it, not where it keeps the data. A European subsidiary with a US parent remains within reach.
The managed service infrastructure runs on European-owned providers with data centres in the European Union. The condition is verifiable in the public subprocessor inventory, not solely in a contractual statement.
Opendome is deployed in three ways. In two of them the data resides in the customer’s infrastructure; in the third, in Opendome’s. European jurisdiction and the storage format are common to all three.
| Model | Who operates | Where the data resides | Opendome's GDPR role |
|---|---|---|---|
| Managed | Opendome, end to end | Opendome infrastructure, on European providers with data centres in the EU | Processor |
| BYOC Bring your own cloud |
Opendome operates the software; the customer supplies the infrastructure | The customer's infrastructure, in the location they determine | Processor |
| Self-host | The customer or an integrator they appoint | The operator's infrastructure; Opendome takes no part in operations | Not applicable |
In all three models the customer is the data controller. Opendome does not determine purposes and does not use the data for its own ends. In the managed model and in BYOC it acts as processor, under a processing agreement pursuant to article 28 GDPR; in self-host it does not process customer data.
How controls are split between the parties in each model is set out in Shared responsibility.
A contractual commitment not to access the data, not to monetise it and to respect the customer’s jurisdiction can only be checked indirectly. An open source system on standard formats can be checked directly. Opendome takes the second route for auditability, not only for cost.
The customer, their auditor or the regulator can examine how data is processed, where it is stored and what information leaves the infrastructure, without relying on the provider’s word.
Data is stored in Apache Iceberg and Lance, readable by any compatible engine in the ecosystem. On a change of provider it remains accessible with no prior conversion.
Certain architectures on the market expose the organisation’s data through APIs and charge for each operation. At Opendome there is no metering or charge for agents accessing customer data.
Each of the statements above, mapped to the source where it can be checked.
A working session to determine the applicable deployment model, the resulting residency of each data set, and the documentation needed for your internal approval.
Request a technical meetingSix European regulations apply to an enterprise AI deployment, and most of them ask for the same things in different words. This manual reduces them to four common questions and explains what each one requires, of whom and from when. It covers the four layers of data sovereignty. A shared reference for executives, business, technology and legal teams.
Descargar el whitepaperBusiness memory already exists: it lives in the data. The only memory that should belong to the agent is procedural.
guardrails vs data security"The agent has guardrails, so the data is protected." That sentence conflates two things operating on different planes. Guardrails watch what the agent says; they don't govern what it accesses. Only structural, reproducible access control protects the data — because it isn't statistical: it's deterministic.
Enterprise AI strategyThere's no enterprise AI strategy without centralizing, relating, describing, and protecting the data. They aren't four good ideas to choose among. They're four conditions, and missing one invalidates the rest.
Ver todos los recursos