FinOps and your organisation’s AI policy

Your cloud data.
Your processing choices.

Separate application hosting from processing by AI and integrations. Review the data sent and the available framework, then let the owner confirm the policy suited to your organisation, including keeping AI disabled.

  • Policy set by the owner
  • Minimised data sent to AI
  • No fallback to Standard AI
01 — UnderstandDocumented processing boundaries · This illustration does not trigger an AI call
Application hosting
France
Enabling AI
Your choice
Sovereign-mode requests*
EU + EFTA

EU + EFTA applies to Sovereign-mode requests, subject to availability. Technical metadata may be processed elsewhere. Application hosting, AI and integrations have distinct scopes.

01Understand

Hosting is a location.
Processing has a scope.

The application, identity service, database and its backups are hosted in France. For your review, distinguish that foundation from delivery services, exports, enabled integrations and AI requests, then consult the recipients and conditions associated with each type of processing.

French hosting does not guarantee residency for all processing. The AI service may process technical metadata outside the inference region; analysis inputs are minimised and use temporary references rather than internal names.

Review the data scope
Move2Cloud FinOpsThe actual product.

Actual product screenshot — Data, control and AI choice

Security settings for the fictional NovaRetail Cloud organisation showing three AI policies, with Sovereign AI as the active policy.
Actual application screenshot — AI policy choices for a fictional organisation: Sovereign AI is the active policy.Enlarge to explore

Franceapplication hosting, distinct from other processing

Your choicean AI policy confirmed by the owner

Without AIcosts, budgets and deterministic recommendations remain usable

A decision example

A verifiable AI choice before analysis.

Illustrative workflow using the fictional NovaRetail data above; no customer results or guaranteed savings.

Before
Application hosting and AI request processing have different scopes.
The review
In NovaRetail’s illustrated settings, the owner reviews the three policies and their availability before confirming a choice.
The next action
European mode never falls back to Standard AI. With AI disabled, costs, budgets and deterministic recommendations remain available.

Processing locations

French hosting has a defined scope.

The application, identity service, database and its backups are hosted by Scalingo in France. Emails, reports, exports and a backup copy use Scaleway in Paris.

Distinguish hosting from AI

The application’s location does not describe every processing activity. AI requests follow the selected mode, and the AI service may handle technical metadata outside the inference region.

Include enabled services

Integrations enabled by your organisation have their own processing scope. Recipients, transfers, retention periods and rights are detailed in the privacy and contractual documentation.

This page is neither a compliance certification nor a guarantee that every processing activity stays in the European Union.

Review recipients and retention

02Evaluate

Choose the framework.
Know its limits.

Compare AI disabled, Sovereign AI and Standard AI by examining what may be sent, where requests are processed and whether the mode is available. The same minimisation principle applies to both AI modes; enabling them requires an explicit choice by the owner.

Sovereign mode processes requests in the EU and EFTA, subject to availability, without fallback to Standard AI. Technical metadata may be processed elsewhere; the modes do not guarantee identical output quality.

Compare AI policies

Three explicit modes

Choose the framework that fits your organisation.

🇪🇺 Sovereign AI

Requests are processed in the European Union and EFTA: Norway, Iceland, Liechtenstein and Switzerland. The service’s technical metadata, such as account, billing and usage statistics, may be processed outside that region. Switching to Standard AI as a fallback is prohibited.

🇺🇸 Standard AI

Optional AI features use minimised data. Processing may take place in the United States under the transfer framework described in the contractual documentation. This mode requires the owner’s explicit choice.

AI disabled

Every new external AI call is blocked, including queued tasks. Existing results remain readable; costs, budgets, alerts, exports and deterministic recommendations remain available.

A request already sent before disabling AI may not be cancellable. Availability does not guarantee identical result quality across modes.

Read the processing scope

03Choose

A shared policy.
An identified owner.

The owner confirms the policy in the organisation’s security settings. Changes record their author and date, and tasks recheck the policy when they run. Review cloud permissions and the integrations you choose to enable separately from this AI setting.

Disabling AI blocks new external calls, including queued tasks. A request already sent may not be cancellable; existing results remain available and FinOps features that do not require AI remain usable.

Understand data control

An organisation-wide choice

One policy, with recorded changes.

A clear decision maker

The owner confirms the policy in Organisation settings → Security → Artificial intelligence. Organisation administrators can view the setting without changing it.

A history of decisions

Changes record the author, date, previous policy and new policy. Tasks read the policy again when executing, so a previous choice does not authorise new calls.

Enforced boundaries

An unavailable European mode does not trigger a switch to Standard AI. Activation depends on availability controls and a current configuration assessment.

Understand control of your data

A plan for your scope

From your first cloud to teamwork.

Review your AI policy and processing requirements with our team first. AI credits are used only if your organisation enables AI.

Standard

€69 excl. VAT / month

For a small team’s cost and budget tracking.

  • Up to €5,000 in tracked cloud spend per month
  • 3 connections · 2 members · 5 dashboards
  • 800 AI credits / month, if AI is enabled
  • Budgets, alerts and direct cost allocation

Pro

€249 excl. VAT / month

For tracked actions, measured savings and responsibilities.

  • Up to €20,000 in tracked cloud spend per month
  • 10 connections · 5 members · 20 dashboards
  • 1,500 AI credits / month, if AI is enabled
  • Savings Tracker, Slack, Teams, Jira, GitHub and GitLab

Starter — free, up to €1,000 tracked spend / month and 1 connection. Without a subscription after the trial, the organisation continues on Starter, within its limits.

Enterprise — on quote, for a tailored scope and governance. Discuss your requirements →

Compare all plans →

An explicit framework

The detail, when you need it.

Data sent to AI

Useful analysis facts, with temporary references.

Both AI modes apply the same minimisation. Real names are used locally when displaying results.

What may be sent

Cost amounts and periods, service categories, normalised regions, budgets and structured facts needed for analysis. Projects and resources use temporary references; results are matched to your data inside the application.

What is removed

Account numbers, email addresses, organisation, cloud account and member names, and access credentials. Resource names, tags and free-form metadata are not sent as entered.

Free-form descriptions

Dashboard descriptions are interpreted locally to avoid sending text that may contain sensitive data. With AI disabled, composition uses local rules.

Your resources’ contents

Connectors read costs and authorised inventory. They do not read the contents of files, databases or messages to send them to AI.

Check data read and retained

Your security review

Evaluate the framework before connecting your data.

Review the commitments

Read the Security and data part and the privacy policy, then identify the processing and integrations relevant to your scope.

Choose the policy

The owner confirms the mode in the organisation. Available modes are shown in the setting; you can keep AI disabled.

Control access

Grant useful permissions at each provider and revoke them when needed. Cloud permissions and the AI policy are managed separately.

Read the data processing agreement

Control stays in your hands.

Your data.
Your rules.

Minimised data.

Analysis uses structured facts and temporary references, without sending internal names or access credentials as provided.

No Standard fallback.

An unavailable Sovereign mode does not switch to Standard AI. The affected features depend on verified availability.

A documented framework.

Recipients, transfers and retention are detailed in the contractual documentation. This scope is neither a certification nor a guarantee of residency for all processing.

Set your processing requirements before your first analysis.

Your requirements first.
Your policy next.

Explore the platform and its documentation. Your organisation’s owner chooses whether to use AI within an available framework that fits your policy.

  1. 01

    Review the commitments.

    Read Security and data, the privacy policy and the data processing agreement.

  2. 02

    Choose the policy.

    The owner confirms an available mode or keeps AI disabled.

  3. 03

    Define cloud access.

    Grant the permissions needed and revoke them when appropriate.

Before you begin

The right
questions.

Does Sovereign AI mean everything stays strictly inside the EU?

No. The request-processing scope covers the European Union and EFTA, including Switzerland, Norway, Iceland and Liechtenstein. The AI service’s technical metadata may be handled outside that region. Enabled integrations also have their own processing locations.

Who can change the AI policy?

The organisation’s owner can confirm and change it. An organisation administrator can view it without changing it. Other roles follow their existing access permissions.

What happens if European mode is unavailable?

The platform does not switch to Standard AI. Activation requires valid configuration and assessment. Unavailability does not change the organisation’s choice automatically; features that do not use AI remain usable.

What happens when AI is disabled?

New external calls are blocked, including queued tasks. Existing results remain readable without updates. A request already sent may not be cancellable. Costs, budgets, alerts, exports and deterministic recommendations remain available.

Is result quality identical in both modes?

We do not promise identical quality. Both modes use the same minimisation framework, and results should be checked before any decision. Recommendations and estimated savings require your team’s validation.

Where can I find contractual information about service providers?

The privacy policy describes recipients, transfers and retention; the data processing agreement adds the contractual commitments. Our team can discuss the questions relevant to your security review.