Control infrastructurefor your AI ambitions.

Delegate valuable work to AI.
Retain authority over the outcome.

Read Oorbital without 3D.

Your browser does not have WebGL2 available. All the text, diagnostics and contact options are available in reading view.

Oorbital

Your next step.

This destination is reserved for the next build.

Intelligence · Example

Less repeated work.
The same care.

Oorbi keeps researching answers already approved by your team. Reusing those answers is an opportunity to reduce repeated work.

The suggested change leaves the £200 refund limit in place. It still needs testing before release.

A local example, not a measured performance or savings claim.

The AI Abundance Report

Preview from the supplied report introduction.

Read more deeply.
Decide more clearly.

Ideas for the people choosing what AI should do next. Open a book, select text and keep your own notes.

All Insights rooms
Choose a desk, or scroll to the reading list.
Insights
Back to the office
Eight questions. A clearer next step.
oorbital

Take your time. Oorbi has the next question.

Your work, saved

Your diagnostic results.

Developer quick start

Keep your stack.
Add an execution boundary.

RESEARCH / REPORT

Shadow AI

The AI your organisation already runs but does not yet manage

Cover of the Oorbital Shadow AI report: an office where AI is already working, with untracked accounts, unapproved tools and hidden automations around it.
29 pages · PDF

Most organisations already have AI taking part in work outside the visibility, ownership or authority built to manage it. This report sets out where Shadow AI arises, how to discover the work rather than only the applications, and where control must move from assistance to delegated execution. It closes with a practical executive checklist and three engagement pathways.

Three signals, three different measures

31%
Undeclared useUK GenAI users reporting use without employer knowledge.Deloitte, 2026 ↗
78%
Bring-your-own AIAI users who brought their own tools to work.Microsoft and LinkedIn, 2024 ↗
47%
Personal AI accountsObserved GenAI users using personal applications or accounts.Netskope, 2026 ↗

These figures measure different populations and different behaviours. They must not be combined into one prevalence estimate. Read each as a signal that unmanaged use is common, not as the rate in your organisation.

What Shadow AI means

Shadow AI is AI participation in organisational work outside the proportionate visibility, authorisation, accountability or controls required for that work. Eight overlapping forms help you look beyond the application list. This is an analytical framework, not a formal classification standard.

  • Shadow toolsUnreviewed applications.
  • Shadow accountsUnmanaged identities or tenants.
  • Shadow dataUnauthorised information flows.
  • Shadow workflowsUndeclared AI participation in a process.
  • Shadow automationUncontrolled scripts or integrations.
  • Shadow modelsUnregistered model hosting or adaptation.
  • Shadow agentsUnmanaged tools, memory or authority.
  • Shadow decisionsUndeclared material influence on decisions.

The execution boundary

The control requirement changes when AI moves from assistance to delegated execution. The same interface can conceal a very different level of authority.

AI assistance

AI proposes. A person assesses and acts.

Delegated execution

AI proposes or selects. Authority is checked. The system acts.

Technical access does not establish business authority. Controls must precede the consequence they govern, including disclosure of data.

The operating response

  1. Discover
  2. Authorise
  3. Control
  4. Evidence
  5. Improve

Find systems, accounts, workflows and information flows. Assign owners and define permitted, conditional and prohibited actions. Apply checks before consequential actions execute, link decisions to outcomes, and use the evidence to increase useful, trusted autonomy.

From assessing to acting

  1. Assess your exposureTake the diagnostic to see where visibility, information, authority, control and enablement are weakest.
  2. Map one workflowExplain the information, authority and consequence path for one material workflow.
  3. Start a Shadow AI DiscoveryA bounded engagement with Oorbital to establish where AI is operating, which exposure matters and what should happen next.

Get the full Shadow AI report

A practical executive guide to discovering Shadow AI, prioritising exposure and moving from unmanaged adoption to controlled AI work.

We use your details to send the report and to follow up if you ask us to.

Engagement scope, available tools and integration coverage are agreed for the customer's environment. These services do not guarantee complete discovery, legal compliance or the prevention of every incident.

Oorbital report · Shadow AIOORBITAL · INSIGHTS
Oorbital

Share a private copy

Share your results.

Choose who receives your results. Nothing is published online.

Open an email draft ↗

Your email app handles sending. This page does not send messages.