Privacy, data quality and governance

A processing agreement cannot control a moving pipeline.
The running operation has to.

A standard consent configuration cannot represent every person, purpose, channel and changing context. Datastreams translates the organisation's approved business and policy rules into the way the runtime handles each individual interaction.

Map policy to runtime Explore the trust model

Runtime controls support accountability and evidence. They do not determine legal grounds or guarantee compliance on behalf of the accountable organisation.

Paper agreement versus runtime control

Static documentation records intent. Live data needs continuous enforcement.

A processing agreement remains necessary, but data sources, context, people, providers and destinations keep changing. The control gap appears when nobody can show that the running pipeline still behaves according to the approved agreement.

The documented agreement

Purpose, parties, categories and instructions are written down.

The document establishes responsibilities and expected processing. By itself, it cannot validate an event, refuse an unauthorised action, apply a changed rule or prove which code path produced an outcome.

The running agreement

The approved conditions are evaluated with every relevant operation.

The active version controls what may be received, combined, used, shared or retained. Runtime evidence records which context, quality check, rule and authority produced the resulting action.

Individual by design

The rule must be resolved for this person, purpose and moment.

Standards and legislation establish requirements. They do not produce one universal permission that is correct for every interaction. The runtime must evaluate the organisation's rules against the current individual context before data is used or an action follows.

Local government services

The permitted data and action can depend on the public task, the resident's request, delegated authority, case status and the channel being used.

Healthcare interactions

Treatment context, professional role, urgency, purpose and the individual's choices can change what may be viewed, combined or shared.

Age and identity proofs

A service may need proof that a condition is met, such as being over a required age, without receiving the person's full identity or date of birth.

AI-assisted decisions

The relevant data, explanation, human oversight and permitted action depend on the use case, the person affected and the role of the organisation.

individual + channel + purpose + authority + current state + policy → permitted data and action

The same business rule can produce a different permitted outcome when the person, relationship, proof, purpose or channel changes. That decision remains inspectable as runtime behaviour rather than being hidden across application code and consultancy implementations.

From principle to runtime condition

Control purpose, quality, authority and change before data becomes an outcome.

A processing definition connects approved business intent to inspectable runtime behaviour. More complex personal, financial or strategic data adds conditions to the same operating model instead of another disconnected control service.

Purpose and grounds

Attach the declared processing purpose and relevant legal basis to the operation that uses the data.

Authority and access

Define which person, application or agent may inspect state, propose a change or execute an action.

Data quality and provenance

Validate structure, source, completeness, timeliness and relevant quality conditions before data influences an outcome.

Change control

Version, test and approve changes to sources, rules, providers and destinations before the running operation changes.

Lifecycle conditions

Make retention, withdrawal, expiry and deletion responsibilities visible in configuration and operations.

Operational evidence

Retain the relevant version, context, policy result, actor, action and destination for review.

Configuration lifecycle

Describe, test, approve, publish and inspect the control.

AI can help translate approved business rules into a proposal, but identity, policy and accountable approval determine what becomes active for every channel.

Describe

State the purpose, parties, data and required outcome.

Validate

Check contracts, references and required conditions.

Test

Use representative events and failure scenarios.

Approve

Record the responsible authority and accepted change.

Observe

Inspect policy results, actions and evidence while it runs.

Shared responsibility

Compliance by design clarifies responsibility; it does not erase it.

Every party retains the decisions and duties that belong to its role.

The organisation

Determines purposes, legal grounds, policies, accountable roles and the authority granted to people and agents.

Datastreams

Provides and services the configured runtime boundary under the agreed deployment and support responsibilities.

Other providers

Identity, trust, AI, infrastructure and destination providers retain their own contractual and regulatory responsibilities.

What becomes easier to answer

Reconstruct less. Inspect more.

When meaning and control are distributed across application code, cloud consoles, contracts and spreadsheets, an audit becomes an archaeological exercise. A declared runtime operation keeps the relevant questions together.

Which version was active?
For which purpose was the data used?
Which authority allowed the action?
Which context and policy result applied?
Where was the outcome delivered?

Regulatory context

Operational control supports obligations that remain with the organisation.

GDPR, the AI Act, sector rules and emerging identity standards overlap differently per use case. Datastreams provides technical and operational controls; applicability, legal interpretation and organisational compliance remain with the relevant controller, processor, provider or deployer.

GDPR accountability

Controllers are responsible for compliance and must be able to demonstrate it. Documentation matters, but so does the technical and organisational operation behind it.

European Data Protection Board

Processor instructions

GDPR Article 28 requires processing by a processor to be governed by a binding contract and documented controller instructions.

GDPR on EUR-Lex

Privacy by design and default

Data protection should be built into systems from the start and remain effective throughout the processing lifecycle.