Home01 Services02 Agent Passport03 Sectors04 Proof05 Pricing06 Self-Assessment07 About08 Contact09

Home / Services

The operating model, installed.

Ambient Agent® is not advisory. It is deployment-led. We design and install the control system that governs how AI is used, who uses it, at what level of capability, and under what conditions it is permitted to scale. Then we hand you the keys.

01 / The system

One system.
Four layers.

From board-level outcomes down to the data and activity it governs. Each layer has defined ownership, metrics and a governance interface. Weakness in one creates exposure across all others.

LAYER 1

Executive Layer — five measurable outcomes

The outcomes AI governance delivers at board and senior leadership level. Every element of the model is designed to deliver against one or more of these five. They are locked. Nothing is added without formal review.

Cost ControlRisk ReductionProductivityScalable AdoptionExec Confidence

The infrastructure that connects executive outcomes to operational activity. This is what makes Ambient Agent an operating system rather than a framework in a document.

Agent PassportGovernancePolicyRiskApprovalsDecision RightsEscalation

The operational components that power the control system. Each engine has defined ownership, metrics and a governance interface. Removing any one engine breaks the system.

E1

Diagnostic

Understand the organisation before configuring the model. Readiness, risk exposure and capability baseline before deployment.

E2

Capability

Baseline and develop the workforce. Role-based programmes built on the Agent Passport framework.

E3

Control

Agent Passport, gates and approvals. Structured access and accountability at every level.

E4

Value

Connect AI activity to business impact. Nothing scales without documented proof of value.

E5

Reporting

Board and exec visibility. Dashboards and reporting built for regulated environments, live from day one.

E6

Comms

Leader-led adoption narrative, aligned to control milestones, not programme milestones.

The foundation of the system. Every input from which the control system operates, observed and governed continuously.

UsersToolsAgentsWorkflowsData sourcesActivity signals
The control system visualAA·DG·SYSTEM
The AI Control System: four layers, one closed loop Executive outcomes, control system core, six engines, and the data and activity layer, connected by a closed evidence loop. LAYER 1 — EXECUTIVE OUTCOMES Cost Control · Risk Reduction · Productivity Scalable Adoption · Exec Confidence What the board sees LAYER 2 — CONTROL SYSTEM CORE Agent Passport · Governance · Policy · Risk Approvals · Decision Rights · Escalation What makes it a system, not a document LAYER 3 — SIX ENGINES Diagnostic · Capability · Control Value · Reporting · Comms What does the work LAYER 4 — DATA AND ACTIVITY Users · Tools · Agents · Workflows Data Sources · Activity Signals Where the evidence comes from THE CLOSED LOOP MEASURE · LEARN · DECIDE · ACT · IMPROVE
The closed loop — the system is continuous, not linear
Measure Learn Decide Act Improve

Activity and performance data feed governance decisions. Decisions are implemented through the engine layer. Effects are measured and fed back. Executive outcomes drive further measurement.

02 / The engines

Six engines. Built to govern.

Each engine has defined ownership, structured inputs and outputs, and a clear role within the operating model. Each is interdependent: weakness in one creates exposure across all others. The model is auditable, sustainable and executive-reportable from day one.

Engine layerAA·DG·ENGINES
Six engines inside the AI Control System A modular 3 by 2 systems graphic showing the six engines: Diagnostic, Capability, Control, Value, Reporting, and Comms. THE SIX ENGINES Six engines. One control system. 01 Diagnostic Understand the organisation before configuring. 02 Capability Baseline the workforce and power the Agent Passport. 03 Control Gated approvals, governance, and capability-based access. 04 Value Convert AI activity into measured business impact. 05 Reporting Board and regulator-facing visibility. 06 Comms Leader-led adoption, not programme noise.

03 / The install

Five stages. One control gate each.

The operating model installs across a five-stage, closed-loop lifecycle. No stage advances without passing its control gate. Progression requires documented evidence of readiness, not intent.

STAGE 01

Mobilise

Week 1

Diagnose the organisation: current state, risk profile, priorities and AI landscape. Executive sponsorship and risk appetite defined. Delivery plan confirmed. Passport framework introduced.

⬖ Control gate

Control profile agreed by executive sponsor.

STAGE 02

Design

Week 2

Baseline capability, configure controls and define decision rights. Roles, capability tiers and platform principles architected. Passport levels assigned to the pilot cohort. L1 baseline set for all staff.

⬖ Control gate

Audience segmentation, pathway model and approval model agreed.

STAGE 03

Prepare

Weeks 3–4

Score and triage use cases, build the reporting MVP and launch the comms cascade. Assessment deployed. Externally accredited programmes activated. L2 Practitioner pathway opened.

⬖ Control gate

Pilot candidates selected and scale criteria agreed.

STAGE 04

Execute

Weeks 5–6

Run governed pilots with control gates, evidence collection and review cycles. Adoption tracked. Value data captured. Reporting live from day one. L3 Builder access governed.

⬖ Control gate

Scale, stop or extend decision made by the steering group.

STAGE 05

Scale

Ongoing

Expand only what has produced evidence. Retire what has not. Governance evolves. New cohorts onboarded through six-month capability reassessment. Client operating the system independently.

⬖ Control gate

Scale approved by the steering group, with evidence.

Each stage is gated. No stage begins without the required artefact and decision from the preceding stage. This is the mechanism that separates a control system from an unstructured consultancy engagement.

04 / Engagement model

Deploy. Prove. Enable. Exit or expand.

This is not a long-term embedded consultancy model. The system is designed to be owned and operated by your team. Ambient Agent exits when the control infrastructure is installed and evidence exists to justify scale.

The three engagements

E·1

AI Control Diagnostic. 2 weeks. The entry point: usage map, risk posture, capability baseline, configured deployment plan. Credited in full against an install within 90 days.

E·2

Control System Install. 6 to 12 weeks. The core product: the full operating model configured to your environment, with capability assessment, governed pilots and a documented scale decision inside.

E·3

Governance Retainer. Ongoing, quarterly. Optional. Reviews, optimisation and governance evolution as regulation develops.