AI Management Business Data Model: AI-Ready Edition

The AI Management Business Data Model: AI-Ready Edition is for CIOs, architects and transformation leads who need to know what information AI management work runs on before an AI agent touches it. It covers 16 subject areas, 116 entities, 830 key attributes and 408 relationships. Every entity and key attribute is classified for sensitivity, with the AI uses and data-quality requirements an agent depends on. It is the data-model part of Business Architecture for AI Management: AI Transformation Blueprint, sold on its own.

What's Inside

The same model comes in three files: the full model in Word, a workbook to filter and extend, and a guide to the scales and the license.

• 16 subject areas and 116 entities, each with a definition, its typical system of record, sensitivity class, AI uses and data-quality requirements
• 830 key attributes with sensitivity and AI relevance, and 408 relationships with cardinality
• Under each entity, the capabilities that need it and the agents that read it, by name and ID
• The Business Data Model (Word): every subject area and entity in full
• The workbook (Excel): subject areas, entities, attributes and relationships in sheets you can filter and extend
• The guide (Word): the scales the model refers to, the data foundation, and the license
• Free corrections, and new versions free for 180 days

Every file names your organization and carries your license ID.

The 16 Subject Areas

The 116 entities are grouped into 16 subject areas. The list gives what each holds and how many entities it has.

• AI Strategy, Portfolio & Value (7 entities): The direction and funding of enterprise AI: the AI strategy and roadmap, maturity assessments, investment cases, funding allocations, portfolio reviews and the benefits that show whether AI pays off.
• AI Use Cases & Inventory (7 entities): Every proposed and running use of AI in the organization: ideas, use cases, feasibility assessments, experiments, and the enterprise inventory consolidated from every function's register, including AI that arrives inside purchased software and AI found in use without registration.
• AI Governance, Policy & Decision Rights (8 entities): How AI is governed across the enterprise: the governance bodies and their charters, the decision rights register, the AI policy, lifecycle and other standards, acceptable use rules, the restricted practices list, time-limited exceptions to standards and the workforce's acknowledgment of policy.
• AI Risk & Impact (8 entities): How AI risk is tiered, assessed and treated: the tiering method, the tier of each use case and agent, the enterprise AI risk appetite, impact assessments and the groups they consider, AI risks, their treatments and recorded acceptance of remaining risk.
• Testing, Validation & Assurance (7 entities): The evidence that models and agents work as intended: evaluation plans and runs outside production, independent validation engagements and their findings, explainability assessments, control self-assessments and the mapping of applicable requirements to AI controls.
• Production Monitoring & Activity Records (6 entities): What models and agents do in production: monitoring thresholds set at validation, observations, drift and harmful output findings, the activity record of every AI action, and periodic performance reviews.
• AI Incidents (6 entities): Harm or near harm from an AI system's behavior: the incident, its classification under the enterprise severity scale, containment actions, the investigation, the facts behind any outside disclosure and the lessons shared across functions.
• Transparency, Explanation & Recourse (9 entities): How people are told about AI and can question it: use notices, system descriptions, the public account of how the organization uses AI, explanations of AI-informed decisions, human review requests, corrections, overrides, concerns and complaints.
• AI Solutions & Model Development (8 entities): How AI solutions and models are designed, sourced, built and documented: solution designs, build, buy or adapt decisions, models, experiments, candidates, instruction versions, output safeguards and model documentation.
• Model Release & Lifecycle (7 entities): How models move into, change in and leave production: model versions, the deployment approvals they rest on, staged releases, changes and urgent fixes, retraining runs, retirements and the processes that depend on each model.
• Agents & Authority (8 entities): AI agents and what they may do: each agent, its versions, written authority limits, named human owner, the identity requested for it, authority changes, release approvals and retirement.
• Agent Operations (8 entities): How agents run day to day: the agent registry, runs, exceptions routed to people, handoffs between agents and to people, interaction rules, performance measures, quality samples and suspensions.
• AI Platforms, Access & Cost (7 entities): What the shared AI platforms must provide and how models are reached and paid for: platform requirements, AI tools, the approved model catalog, usage limits, content safeguard rules, compute allocations and cost records.
• AI Data & Knowledge (8 entities): The data AI is built on and learns from: training data requests to Data and Analytics, references to the data sets it releases, their acceptance, evaluation data sets, knowledge sources and their approval, outcome records and user feedback.
• AI Providers & Partnerships (6 entities): The organizations that supply AI or work with the organization on it: providers, their assessment on AI grounds, AI contract terms, model change notices, performance reviews and partnerships.
• AI Skills & Organization (6 entities): The people side of running AI: AI role definitions, the workforce literacy program, learning assignments, literacy measures, practitioner accreditations and communities of practice.

Who Uses It

Data owners and stewards, data architects, and AI and analytics engineers who need to know what an agent may read, how sensitive it is and how good it must be first; and business architects who link capabilities to data.

How It Fits With the Other Products

The capabilities named under each entity are defined in the AI Management Capability Model: AI-Ready Edition. The agents are specified in Business Architecture for AI Management: AI Transformation Blueprint, which contains this data model and adds the value streams, use cases, controls, an operating model and a roadmap. The AI Management Value Streams are sold on their own too.