CIOPages
Back to Store

Capability Models

AI Management Business Capability Model: AI-Ready Edition

61 capabilities for running AI across the enterprise, each with an AI profile giving target mode, oversight and risk tier.

What this capability model is for

A capability model lists what a business does, separate from who does it and which systems support it. Because it changes less often than the org chart or the application estate, it gives strategy, architecture and finance one map to plan against.

Teams use it to scope a transformation, to check which capabilities each application supports and where two overlap, to assign costs and owners, and to plan the integration after a merger.

About This Capability Model

The AI Management Business Capability Model describes the work of running AI across an enterprise, for CIOs, CTOs and heads of AI who have to set up that work and decide who does it. It names 298 L3 capabilities under 61 L2 capabilities and 12 L1 areas, each defined, with 239 KPIs at L2, covering how uses are chosen, how agents are approved and limited, and how their behavior is watched until they are retired. Every L2 capability carries an AI profile with a target AI mode, an oversight level and a risk tier, and a scoring model ranks them on weights you choose.

About the AI Management Business Capability Model

Most AI inventories list models and tools. This model lists work: choosing AI uses, assigning owners, setting what agents may do, reviewing changes, watching behavior and retiring agents. It is organized as 12 L1 areas, 61 L2 capabilities and 298 L3 capabilities, each with a definition, and the L2 level carries 239 KPIs.

IDs stay fixed between releases. Agent authority management is AM-7.2, so the charter of an AI office, a RACI and a budget line can cite it and mean the same scope.

The AI ratings apply the model to its own function. Each L2 profile states where AI could prepare the work of the people who govern AI, what they keep, the agents and data involved, prerequisites, risks and role changes, with a target AI mode, an oversight level and a risk tier. Each L3 has a work pattern, AI mode, AI fit and risk tier. AM-7.2 is rated Augment with the oversight level Human decides: an assistant drafts an agent's authority limits, and the Head of AI Governance sets them.

What You Get

In the package, ready to adapt:

The structure

  • The AI management capability model: 12 L1, 61 L2 and 298 L3 capabilities, a definition for each one down to L3, and 239 KPIs at the L2 level

The AI layer

  • An AI profile on each L2 capability, setting out opportunities, what stays human, agents, data, KPIs, AI measures, prerequisites, risks and workforce impact
  • Each L3 capability rated for work pattern, target AI mode, AI fit and risk tier
  • A prioritization model that scores and ranks each L2 capability, adjusted for risk, from ratings and weights you set
  • A heatmap of where AI fits across the model
  • An AI readiness assessment of 40 weighted questions

The files

  • A Welcome document to start from, then the capability map deck, standalone capability definitions and KPIs documents, the Excel workbook, the user guide
  • Bonus files in the same download: an AI management glossary, A Practical Guide to Business Architecture, a capability modeling overview, an AI-ready capability profile template and capability relationship mapping templates
  • Corrections at no charge, and any new version released in your first 180 days

Each file carries your organization's name and your license ID. The ratings are set for a typical large enterprise, as a first draft you adjust to your own.

Why It Matters

A CIO is often the executive asked to account for every AI system in the company, including the ones business units bought themselves. Without a defined body of AI management work, that account is assembled from procurement records, cloud bills and committee minutes, and the gaps show up later as agents nobody owns or changes nobody reviewed.

The model gives that work a structure the CIO can staff and fund. Each capability can be given an owner, a team and a KPI, and the split between the AI office, security, risk, data and the business can be drawn on one map. The AI profiles add a second decision: where AI can prepare governance work, such as drafting an agent's limits, and where a named person must decide.

Who Works With It

CIOs and CTOs. Decide where AI management sits, who owns each capability and what it costs.

Chief AI officers and heads of AI. Scope the AI office and set its KPIs from the L2 capabilities.

Heads of AI governance. Use the profiles to choose which governance tasks an assistant may prepare and which stay entirely with people.

Enterprise architects. Link AI management capabilities to the enterprise model and to the agents running in other functions.

Risk, security, privacy and data leads. Agree which AI management capabilities they own and which they share with the AI office.

Where It Is Used

Deciding where the AI office reports. Mark the capabilities the AI office would run, those that stay with IT, security, risk and data, and those the business keeps. Comparing two or three reporting options on that map shows which leaves the fewest capabilities split between owners.

Bringing business-bought AI into one process. Place every AI tool a business unit bought on the capabilities that should have governed it, such as intake, approval and monitoring. The capabilities each purchase bypassed show where the process needs a route for buying outside IT.

Costing the work of governing AI. Tag the cost of review, monitoring and inventory work to L2 IDs. The executive team sees what governing AI costs per capability, separately from the cost of the AI itself.

Scoping the first AI governance roles. Write the scope of the first roles in the AI office from the capabilities they will own and their KPIs. The profiles show which parts of each role an assistant can prepare.

Reporting AI oversight to the board. Report the state of AI management by L1 area, with the readiness score and the KPIs behind it. Board members see which parts of oversight are in place and which are still planned.

Setting one default for agent authority. Adopt the AM-7.2 arrangement, in which an assistant drafts and a named person sets an agent's limits, as the default for every enterprise agent. Record each exception against its capability ID.

A Blank Slate, Consultants, or a Reference Model

A CIO setting up AI management can define the work internally, have a firm define it or adapt this model.

  • Building it from a blank slate. The heads of AI, security, risk and data agree the work between them and rate each capability for AI. The function is new in most enterprises, so there is little internal material to start from, and the argument about ownership tends to begin before the list of work exists.
  • Hiring consultants to build it. A firm brings its own model of an AI office and fills it in through workshops. The structure reflects the firm's methods, and unless the scoring model is handed over, the ranking cannot be rerun after the engagement ends.
  • Starting from this model. The team starts from 298 defined capabilities, rated and scored, and spends its time on ownership: which capabilities the AI office runs and which stay with existing teams.

Advisers are still useful for the decisions that depend on your company: where the AI office reports, how it works with risk and security, and the weights in the scoring model. The ratings describe a typical large enterprise.

First Steps

A CIO can set up the work of AI management from this model in these steps:

  1. Open the Welcome document first, then the user guide, to see how the files connect.
  2. Walk the capability map deck with the heads of AI, security, risk, privacy and data, and record who owns each L1 area today.
  3. Mark the capabilities nobody owns and decide where each will sit.
  4. Set the scoring weights, change the ratings where they do not fit and run the readiness assessment.
  5. Turn the top-ranked capabilities into the first year's plan for the AI office, with an owner and KPIs for each.

The Free Sample

The free sample is 0.65% of the package. It follows a single thread through it: agent authority management. These fields are from the Agent Authority Management (AM-7.2) AI profile:

  • Where AI can help: Draft an agent's written authority limits from its approved design and the agent authority standard: what it may read, propose, change and send, and what it must route to a person
  • The decision people keep: The Head of AI Governance sets or changes the authority limits of an enterprise agent on what the Agent Authority Review Assistant prepares, after hearing the Enterprise AI Product Owner, the Agent Platform Lead, the Security Architect (cybersecurity) and the Privacy Lead
  • Kind of work: Rules-based decision, Judgment-based decision, Analysis & insight
  • Target mode: Augment
  • Oversight: Human decides
  • Scores for AI fit, value, feasibility and data readiness: 3, 5, 3 and 2 (scale 1–5)
  • Risk: High
  • Roadmap: Wave 2

Moving Up Later

Upgrade to the AI Management Business Architecture Toolkit: AI Transformation Blueprint within 90 days and the full price of this edition is credited toward it. Send your order number to support@ciopages.com and we reply with a one-time checkout code.

Questions Before You Buy

Which edition do I need?

The AI-Ready Capability Model is enough if you need the capability map, an AI profile for every L2 capability, prioritization, the heatmap and the readiness assessment. The AI Transformation Blueprint holds all of that and adds value streams redesigned for AI, the business data model, agent specifications, use cases, responsible AI principles and controls, the operating model and the roadmap, a leadership presentation template, and bonus files on roles, operating model archetypes and transformation.

Can we use the content in our own AI tools?

Yes, for internal use. Under the Enterprise license you can put the content into your internal tools, your own assistants and agents among them. Neither the Enterprise nor the Consultancy license lets you use it to train, fine-tune or prompt AI systems for third parties.

Are AI laws and regulations covered?

No, on purpose. The package is business architecture: it sets out what AI management does and where AI fits, as capabilities, value streams, data and an AI layer, and that structure holds whichever laws apply where you operate. The risk tiers, principles and controls are durable practices, and your compliance team maps them to your own jurisdictions.

What changes after 180 days?

Free new versions end; the license does not. You can use the files with no end date, and corrections stay free for as long as you do. Any new version released in the first 180 days after purchase is free, and we email you a link to download it. Later versions are offered to you at 25% off their price, with nothing to renew.

Can I start with the Capability Model and upgrade?

Yes. Upgrade within 90 days and everything you paid for the Capability Model is credited toward the Blueprint. Send your order number to support@ciopages.com and we reply with a one-time checkout code.

Why not have ChatGPT or another AI assistant write this?

An assistant can outline what an AI office does, and for a first conversation that outline is useful. Deciding who owns agent approval, monitoring and retirement across a large enterprise needs a list that holds still. This model gives 298 L3 capabilities stable IDs, validates the links between capabilities, ratings and the readiness questions before release, rates every L2 capability for AI mode, oversight and risk tier, and ranks them with weights you set. A generated outline changes with each prompt and carries no IDs to assign owners against.

Asking a chat assistantIn this package
Structure and IDsA list that sounds right and differs on every runStable IDs on 12 L1, 61 L2 and 298 L3 capabilities, each defined down to L3
Links between partsNobody checks themLinks between capabilities, ratings and the readiness assessment are all checked before each release
AI placementIdeas you check on your ownEach L2 capability profiled for AI mode, human oversight, opportunities, what stays human, data, KPIs, prerequisites and risks
RiskGeneral warningsEvery L2 capability carries a risk tier, and those that inform a decision about a person are flagged
PrioritiesNothing to score againstScoring from ratings and weights you set, and a readiness assessment of 40 questions
Over timeChanges with the prompt and the model versionA description of what AI management does that depends on no tool, vendor or law, corrected at no charge, with new versions free for 180 days

Related Capability Models

Keep reading

Checklist
Board Presentation Readiness Checklist

Ensure your IT and security presentations are board-ready.

Decision tool
Technical Debt Prioritization Tool

Rank your technical debt items by business impact, effort, and risk in a quadrant view.

Customer Service and Experience Business Capability Model: AI-Ready Edition