AMBIENT AGENT
Self-AssessmentBook an introductionIntroduction

AI arrived before the controls did.

Your people are already using AI at work. Much of it was never formally approved, and often nobody owns it. Ambient Agent® installs the controls that close this gap, in weeks, then hands them to your team.

Book an introductionTake the Self-Assessment

Nothing scales until it passes the gate.

Cyber EssentialsCertified by IASME. Valid to August 2027.
Public sectorAppointed to the RM6200 Artificial Intelligence DPS.
MicrosoftMember of the Microsoft AI Cloud Partner Program.
Registered trade marksAmbient Agent® UK00004193198. Agent Passport® UK00004408199.

The problem

AI is in use faster than anyone can approve it.

This is the ordinary position, not a failing. The numbers below come from published UK research and from one real organisation.

71%

of UK employees have used AI tools at work that their employer had not approved. More than half still do so every week.

Microsoft UK, research by Censuswide, October 2025. Survey of 2,003 UK employees. Source

39%

of UK workers surveyed have put company information, such as financial, sales or customer data, into a public AI tool.

KPMG and the University of Melbourne, Trust, attitudes and use of AI, 2025. UK sample of 1,029 people. Source

24%

of UK businesses that use AI, or are considering it, have cyber security measures in place to manage the risks from AI.

Department for Science, Innovation and Technology, Cyber Security Breaches Survey 2025/2026, April 2026. Official statistics. Source

One organisation counted its AI agents. It found 1,224, and 109 of the live ones had nobody responsible for them.

An AI agent is an AI tool set up to carry out a task on someone’s behalf. Each square is one agent. Each amber square had no recorded owner, so nobody would have noticed if it went wrong. Roughly nine in ten records held no sharing data, so nobody could show who could see what.

This is rarely deliberate. Capable tools arrive inside licences people already hold, faster than any approval process was built to handle.

By early October 2026, every live agent had a named owner. Read how it was done

One square for each AI agent found: 1,224No recorded owner: 109

What we install

One control system. Four things you can finally show.

We install the rules, approvals and records around the AI your people already use. There is no new platform to buy. It runs on the tools you already have.

1

See what AI is in use

Every AI tool and agent goes on one register, with a named owner beside it. You stop guessing how much there is.

AI registerSample data
128AI tools and agents found
119with a named owner
9with no owner yet
14waiting for approval
AI in useNamed ownerRiskStatus
Claims letter drafting assistantHead of ClaimsHighApproved
HR policy helperPeople team leadLimitedApproved
Meeting summariser built by a teamNo ownerNot ratedFound, not approved
Chat feature inside supplier softwareProcurementLimitedIn review
Illustrative view with sample data. Built in the tools you already run.
2

Decide who can do what

The Agent Passport® gives each person a level based on what they have shown they can do. Access follows the level.

What the Agent Passport® is

Agent Passport® levels across the workforceSample data
Level 1: Consumer72%
Level 2: Practitioner24%
Level 3: Builder4%
LevelWhat it allows
1 ConsumerUses approved AI and the agents shared with them. Cannot build.
2 PractitionerBuilds simple agents, with approval.
3 BuilderBuilds advanced agents, with full sign-off.
Illustrative view with sample data. Built in the tools you already run.
3

Approve on evidence

Before a use of AI is rolled out, it passes the control gate. Three checks, a named approver and a dated record.

How the control gate works

Control gate: approval recordSample data
Use of AI
Claims letter drafting
Business Owner
Head of Claims
Risk rating
High
Decision
Cleared to roll out
Is this person passported?The builder holds Level 3. The users hold Level 1 or above.
Is there evidence?Use, data and risk are on record, with trial results.
Has someone accountable approved it?Signed off by the Governance Approver, with the date recorded.
Illustrative view with sample data. Built in the tools you already run.
4

Show the board and the regulator

Your leaders see how much AI is in use, who owns it and what has been approved. If a regulator asks, the answer is already written down.

Board report: AI under controlQuarter 3, sample data
128AI tools and agents on the register
93%with a named owner
14approvals this quarter
3stopped at the gate
Uses with evidence on file88%
Staff holding a level76%
High-risk uses reviewed100%
Uses with measured value41%
Illustrative view with sample data. Built in the tools you already run.

How you start

Every step has a published price.

Three pieces of paid work, each complete in itself. When the system is running, your own team owns it.

  1. Self-Assessment

    Ten questions on this site. About five minutes.

    No charge

  2. Introduction

    30 minutes. A conversation about where you are, then a look at a sample of the work.

    No charge

  3. Your team runs it

    You own the system. A Governance Retainer is there if you want it.

    Optional: from £10,000per month, plus VAT
See full pricingHow the system is installed

Proof

Proven in two real deployments.

Brent Phillips, founder of Ambient Agent

Founder Brent Phillips led AI governance and enablement inside two large regulated organisations. Both organisations have been anonymised.

One found 1,224 AI agents and gave every live agent an owner. The other went from design to live governance in six weeks.

Read the case studies
1,288AI agents on one register, every one reviewed
109 to 0Live agents with no owner, now none
18.6%More complaints handled, no extra headcount
£230k+Saving identified in the first deployment

The rules

The obligations are already running.

Parts of the EU AI Act already apply, and it can reach UK organisations. UK regulators are applying the frameworks they already have. Our work is designed around the EU AI Act, the UK Government’s principles for AI regulation, the NIST AI Risk Management Framework, ISO/IEC 42001, and what the FCA and the ICO expect.

Reviewed monthly. Last reviewed 11 October 2026.

  1. AI literacy obligationIn force
  2. Transparency obligationsIn force
  3. Stand-alone high-risk obligationsTo come
  4. AI embedded in regulated productsTo come
See the Regulatory Clock

Who it is for

Start from where you sit.

You do not need to be an AI specialist to start this conversation.

I lead the organisation

You have been asked what is being done about AI, and you want an answer you can stand behind.

For leaders and boards

I run IT or technology

You can see AI use growing and want it under control before something goes wrong.

For IT and technology

I own risk or compliance

You already carry the obligations. AI now has to fit inside them, with evidence.

For risk and compliance

See a sample before you decide anything.

An introduction takes 30 minutes and there is no charge. We talk about where you are, then show you a sample of the work.

Book an introductionTake the Self-Assessment