Envisago
AIVOM™ Architecture

The architecture behind AIVOM™

AIVOM™ is a maintained system of four connected dimensions, twenty sub-dimensions and over ninety components. It is designed to show where an operation stands with AI, what is constraining operating impact and where to focus next.

Download the AIVOM™ Architecture Start your free Briefing
The AIVOM Architecture document, cover page with the four-dimensional AIVOM Loop
Purpose

What this document explains

AI changes more than the tools. It changes how work flows, where judgement sits, what people need to be able to do, how quality is managed and how performance is evidenced. AIVOM™ brings those conditions into one connected operating loop so they can be understood and acted on together.

The existing AIVOM™ page explains what the system is and why it matters. This page, and the public architecture document it carries, explain how the system is structured, how a reading is produced and why the method can be trusted.

Architecture at a glance

One loop. Four dimensions. Twenty sub-dimensions. Over ninety components.

The loop and four dimensions provide the stable core. The twenty sub-dimensions, presented across the products as areas of operation, organise the operating conditions inside each dimension. The component layer provides the depth needed to interpret what each state means, what may be constraining impact and what progressing requires.

The AIVOM continuous loop connecting Value, Design, Capability and Performance

Value

The value being sought, how it aligns to direction and whether it is owned and evidenced.

  • Strategic alignment
  • Value goals
  • Investment and cost visibility
  • Accountability
  • Value evidence

Design

How the operation works with AI, including workflows, information, quality and governance.

  • Workflow design
  • Data and systems readiness
  • Operational knowledge readiness
  • Quality architecture
  • Governance

Capability

How roles, responsibilities, judgement and the ability to sustain AI-enabled work are changing.

  • Culture and readiness
  • Training and development
  • Capability distribution
  • Role redesign
  • People performance

Performance

What results are tracked and whether AI is delivering the value the operation set out to create.

  • Financial performance
  • Operational performance tracking
  • Experience tracking
  • Innovation and new value
  • Performance-to-value evidence

The architecture stays recognisable while the reference develops. Sub-dimension names and definitions can be refined as operating practice changes. The component layer is expected to expand more frequently, as new requirements develop.

The reading process

From structured input to an ordered starting point

01 Scope Defines the workflow, function, department or wider operational unit in view.
02 Structured inputs Questions locate the operation against defined positions across the twenty sub-dimensions.
03 Dimension states The areas combine into a state for Value, Design, Capability and Performance.
04 AIVOM™ Loop State The connected system is read through its most constraining dimension rather than averaged upwards.
05 Relational reading The method examines important dependencies and tensions across the loop.
06 Priority areas The most material sub-dimensions are selected in the context of the wider profile. These become the priority areas named in a reading.
07 Binding constraint One place is identified as most holding the loop back and needing to move first.
08 Written interpretation AI explains the governed result in clear, operation-specific language without changing the calculated profile.

The wording can vary; the method beneath it does not. AIVOM™ sets what is examined, what the states mean and how the dimensions are read together. AI personalises the explanation within those boundaries.

State model

Defined states for the dimensions and the loop

A state describes the operating conditions currently in place within the scope being read. It is not a judgement on the organisation as a whole.

These are not maturity stages, and there is no score to climb. A state is read within its scope, the loop is read through its most constraining dimension, and the reference itself keeps moving.

Value clarity

Unclear AI activity is not anchored to clearly defined value goals, ownership or evidence.
Directional The intended value is broadly understood, but goals, ownership, cost visibility or evidence remain incomplete.
Defined Specific value goals, ownership and an approach to evidence are in place, though not yet consistently aligned across the scope.
Aligned AI decisions, investment and performance evidence are consistently connected to agreed enterprise value.

Operating design

Unadapted AI has entered the operation without materially changing workflows, information, controls or decision points.
Reactive Local adjustments are being made, but the operating design remains uneven and largely responds to issues as they arise.
Redesigned Workflows, data, knowledge, quality and governance have been deliberately redesigned to support AI-enabled operation.
AI-native Work is designed around human and AI execution from the outset, with the necessary controls and improvement mechanisms built in.

Capability readiness

At risk Roles, skills, judgement and readiness are insufficient to operate AI reliably across the scope.
Forming Capability is developing, but remains concentrated, inconsistent or dependent on individual initiative.
Enabled Roles, skills, authority and performance expectations deliberately support AI-enabled work.
Distributed Capability is embedded broadly enough for the operation to sustain, challenge and improve how it works with AI.

Performance evidence

Unproven There is insufficient reliable evidence of AI’s effect on operational performance or value.
Anecdotal Positive effects may be visible, but evidence is selective, informal or inconsistently measured.
Measured Defined measures track AI’s operational effects and connect them to the outcomes being pursued.
Proven Sustained evidence shows that AI is producing the intended performance and value, and that evidence informs further decisions.

The AIVOM™ Loop State

The system-level reading is gated rather than averaged: a connected operating loop cannot be stronger than the dimension currently constraining it.

Fragmented AI activity is present, but the four dimensions are not yet working as a connected operating loop.
Emerging The foundations are beginning to form, but the connection between AI activity and operating impact remains uneven.
Connecting The dimensions are linking up, though the loop is not yet consistently self-reinforcing.
Compounding Value, Design, Capability and Performance reinforce one another so impact can become deliberate, evidenced, repeatable and scalable.
System logic

The operation is read as a connected whole

AIVOM™ does not treat Value, Design, Capability and Performance as independent workstreams. A constraint in one changes what is possible in the others. The relational reading examines dependencies, tensions and reinforcing conditions across the profile.

The purpose is not to improve everything at once. AIVOM™ identifies the priority areas that matter most in context, then identifies the binding constraint: the one condition most holding the loop back and needing to move first.

Value → Design Clear value goals give operating design a target.
Design → Capability A changed workflow alters where judgement sits, what roles must do and what capability needs to be distributed.
Capability → Performance A sound design cannot produce reliable performance if the people running it cannot sustain, challenge or improve it.
Performance → Value Performance evidence shows whether the intended value is being created and informs the next decision.
Comparison and benchmarking

Benchmarking designed for continuous improvement

AIVOM™ benchmarks the operation against maintained operating states, its own earlier baseline and the value and performance goals it set out to achieve. The purpose is to show whether the conditions around AI are developing, whether the loop is strengthening and whether AI is producing the intended operating impact.

The emphasis is on internal continuous improvement rather than broad peer comparison. AI is placed differently in every operation: within different workflows, at different levels of autonomy and with different data, quality, governance and human judgement requirements. Headline comparison between organisations can therefore obscure more than it reveals unless that context is properly controlled.

Current position

Where the operation stands against the latest AIVOM™ states.

Progress from baseline

What has moved since the earlier reading and whether the priorities have changed.

Performance against goals

Whether AI is producing the value and operating impact the operation set out to achieve.

External data may add useful context as AIVOM™ develops, but it does not replace the operation’s own evidence of progress and impact.

Versioning and governance

A living reference, governed over time

AI-enabled operating practice will continue to move. AIVOM™ is designed to move with it. The four dimensions provide the stable architecture, the twenty sub-dimensions are maintained and the component layer can expand as new operating requirements emerge.

Each new reading uses the latest AIVOM™ reference so the operation is read against current expectations. The earlier reading remains its baseline. Where the reference has changed materially, AIVOM™ separates movement in the operation from movement in the reference and uses a valid crosswalk where one is available.

The architecture stays recognisable, the reference keeps moving and the operation keeps moving with it.

Start with your operation

See where your operation stands

The free AI Operating Impact Briefing gives you a personalised, structured read across the AIVOM™ Loop, your priority areas, how they connect and one clear place to begin. It takes about twelve minutes.

Start your free Briefing Download the AIVOM™ Architecture

For questions about the architecture, contact hello@envisago.com.

Subscribe to The Envisago Compass

Weekly insights on AI Operating Model Design.