TBM for AI
Managing AI economics and value—and using AI to advance Technology Business Management
Artificial intelligence is changing how organizations invest in, deliver, and consume technology. AI capabilities are becoming embedded in enterprise applications, business processes, digital products, and technology operations. Their economics can differ substantially from traditional technology services, with consumption-based pricing, shared infrastructure, variable workloads, specialized labor, and evolving operating requirements.
These characteristics create new management challenges. Technology leaders need to understand what AI costs, what drives consumption, which services and consumers benefit, and whether investments are producing meaningful organizational value. They also need to determine when AI capabilities should be expanded, improved, replaced, or discontinued.
Technology Business Management (TBM) provides a structured approach to these questions by connecting financial, operational, and business information with technology resources, solutions, consumers, and investment decisions.
AI also creates opportunities to improve TBM itself. With appropriate data, controls, and oversight, AI can assist practitioners with classification, modeling, analysis, forecasting, reporting, and the delivery of insights.
This page explores both relationships: how organizations can use TBM to manage AI investments and operations, and how they can use AI to strengthen the TBM practice.
Understanding the Relationship Between TBM and AI
TBM supports three complementary areas of AI management.
Managing AI spending and consumption
AI costs may originate in cloud infrastructure, model providers, software subscriptions, data platforms, internal labor, external services, and shared enterprise capabilities. Consumption may be measured through tokens, requests, compute hours, licenses, or other units, depending on the technology and commercial arrangement.
TBM helps organizations bring these costs and consumption measures into a consistent management context. By connecting financial records and operational usage with technology services and consumers, leaders can understand spending patterns, establish accountability, and identify opportunities to improve resource allocation.
Managing AI investments and organizational value
Knowing what AI costs is necessary, but insufficient for determining whether an investment is worthwhile.
AI initiatives may be intended to improve productivity, reduce operating costs, enhance experiences, create new capabilities, strengthen decision-making, or support other organizational objectives. Their value depends on what changes as a result of their adoption and whether those changes justify the resources and consequences involved.
TBM connects technology economics with business priorities and relevant performance information, supporting investment evaluation, portfolio governance, and decisions throughout the AI lifecycle.
Using AI to improve TBM
AI can also assist the people responsible for operating the TBM practice.
Potential applications include recommending classifications for financial records, identifying unusual data patterns, explaining cost variances, developing forecasts, preparing scenarios, and helping stakeholders interpret technology information.
These applications require appropriate controls. AI-generated recommendations should remain traceable to governed data, established business rules, and accountable human decisions.
Two directions of AI and TBM
- Use TBM to manage AI: Understand AI resources, spending, consumption, investments, and organizational value.
- Use AI to improve TBM: Strengthen data management, analysis, forecasting, reporting, and practitioner effectiveness.
Both depend on reliable information, clear ownership, and disciplined decision-making.
Managing AI Spending and Consumption
AI spending can be difficult to understand because costs are often distributed across technologies, vendors, organizational budgets, and consumption models.
An AI-enabled application may use a commercial model API, enterprise data services, cloud infrastructure, application integration, security controls, monitoring tools, and specialized personnel. Some costs can be attributed directly to a product or service. Others are shared across multiple workloads or embedded within existing software contracts.
Traditional financial reporting may identify the vendor, account, or cost center responsible for spending without revealing which AI capability consumes the resources or which business activity benefits.
TBM addresses this challenge by connecting financial and operational information through the TBM Model.
Establishing AI cost transparency
A useful starting point is identifying where AI-related costs originate and distinguishing direct costs from shared or embedded costs.
Relevant information may include:
Cloud and model-provider billing, including consumption quantities and charges.
Software subscriptions and AI-related licensing.
Dedicated and shared compute, storage, networking, and data services.
Engineering, development, integration, and operational support costs.
Security, monitoring, evaluation, and governance activities.
Service ownership, application relationships, and business consumers.
Organizations should avoid assuming that all AI costs can be identified through a single vendor, account, or technology category. AI functionality may be embedded in broader services, and some spending may require additional classification or attribution.
Connecting consumption to accountability
Consumption measures become more useful when connected to the services and consumers responsible for demand.
For example, API usage records may identify a model deployment or application, while a service catalog establishes which business capability that application supports. Combining these relationships allows leaders to examine consumption and cost at a level relevant to management decisions.
Shared costs require documented allocation methods. Direct metering is generally preferable where reliable usage data exists. Where direct attribution is unavailable, organizations may use supported consumption drivers or reasonable proxies, provided their assumptions and limitations remain visible.
This supports showback, chargeback where appropriate, budgeting, forecasting, and more informed conversations about demand.
Understanding the Full Economics of AI
AI cost management should extend beyond the price of an individual model request or unit of compute.
Tokens, inference requests, GPU hours, and subscription seats can provide useful measures of consumption. However, these measures do not necessarily represent the full cost of delivering an AI-enabled product or service.
The total economic picture may include development and integration, data preparation, infrastructure, software licensing, evaluation, security, monitoring, operational support, workforce requirements, and ongoing changes to the capability.
Different AI delivery models also create different cost structures. A commercial SaaS capability, an internally developed application using external models, and a self-hosted model may require substantially different forms of investment and operational support.
TBM helps leaders compare these alternatives by relating their costs to the services delivered and the outcomes they are expected to support.
AI unit economics
Unit economics connects the cost of a capability with a meaningful measure of its output or consumption.
Depending on the application, useful measures might include cost per completed business transaction, cost per processed document, cost per supported interaction, cost per active user, or cost per successful workflow.
Technical measures such as cost per thousand tokens or cost per GPU hour remain useful for infrastructure and platform management. However, they should not automatically be treated as measures of business value.
Comparisons must also account for differences in workload complexity, model performance, quality, reliability, and service requirements. A lower unit cost is not necessarily better if the alternative produces poorer results or requires more downstream work.
Total cost of ownership and investment evaluation
Total cost of ownership provides a broader view of the resources needed to establish, operate, maintain, and eventually replace or retire an AI capability.
When connected with business performance information, this supports evaluation of expected and realized benefits, financial trade-offs, and alternative investments.
For a deeper examination, read Beyond Tokenomics: TBM Delivers the Full Economic Picture for Accurate AI ROI.
Connecting AI Investments to Organizational Value
AI initiatives should be evaluated according to the organizational objectives they are intended to support, not simply their technical capabilities or adoption levels.
A successful AI investment may improve Financial Performance, Operational Efficiency, Experience, Innovation, Risk & Compliance, or Sustainability. A single initiative may influence several of these Organizational Value Drivers simultaneously, sometimes in competing ways.
For example, an AI-enabled service may improve employee productivity while increasing technology operating costs. An automation initiative may reduce processing time but introduce new oversight requirements. A new customer-facing capability may improve experience while creating additional operational dependencies.
TBM provides the financial and operational context needed to evaluate these consequences alongside business evidence.
From expected benefits to realized outcomes
An investment proposal may describe expected time savings, reduced costs, improved quality, or increased capacity. These expectations should be connected to measurable changes in the organization.
Time saved does not automatically become a financial benefit. Adoption does not establish effectiveness. Technical accuracy does not necessarily demonstrate an improved customer or employee experience.
Leaders should distinguish operational indicators from realized organizational outcomes and identify the additional evidence needed to establish value.
Managing competing priorities
Investment decisions should account for financial costs, opportunity costs, operational requirements, risk exposure, and expected benefits.
TBM helps leaders compare alternatives, understand resource commitments, and make informed decisions about where AI investments should be directed.
The objective is not to maximize AI adoption. It is to fund and manage AI capabilities that contribute meaningfully to organizational priorities.
Managing the AI Investment Lifecycle
AI investments require continuing management after initial approval.
An initiative that appears promising during experimentation may become uneconomic at scale. A capability that initially delivers value may later be surpassed by another technology, become redundant with existing software, or fail to achieve expected adoption and outcomes.
TBM supports governance across the investment lifecycle.
Experiment and evaluate
During experimentation, leaders need visibility into development costs, resource requirements, intended outcomes, and the evidence needed to justify further investment.
The objective is to learn whether the capability can deliver useful results and whether its likely economics support continued development.
Deploy and scale
As an initiative moves into production, the management focus expands to service ownership, consumption, operational support, security, reliability, financial accountability, and realized business outcomes.
Costs that were insignificant during experimentation may become material at scale. Forecasting should therefore reflect expected usage, architecture, service requirements, and operating responsibilities.
Optimize and improve
Production AI capabilities should be evaluated for opportunities to improve cost, quality, performance, reliability, and business effectiveness.
Optimization may involve changing models, adjusting infrastructure, improving application design, managing demand, or revising operating processes.
Rationalize and retire
Not every AI capability should continue indefinitely.
Organizations may need to consolidate overlapping services, replace expensive implementations, discontinue low-value initiatives, or redirect resources toward more promising opportunities.
These decisions should consider the complete economic and organizational consequences of change, including transition costs, contractual obligations, dependencies, and expected benefits.
Read When AI Doesn’t Deliver: Using TBM to Guide AI Rationalization for a closer examination of these decisions.
Governance as a continuing discipline
Effective governance connects technology investment decisions with financial accountability, operational responsibilities, and business priorities. It should support timely decisions throughout the lifecycle rather than functioning solely as an approval process.
Explore Transforming Tech Governance into Business Decision-Making for related guidance on using TBM to strengthen technology investment governance.
Using AI to Improve the TBM Practice
AI presents opportunities to improve how TBM practitioners collect, prepare, analyze, interpret, and communicate technology information.
Many TBM activities involve large datasets, recurring transformations, complex relationships, and substantial manual investigation. AI can assist with selected tasks, potentially reducing repetitive work and helping practitioners identify issues or opportunities that might otherwise receive less attention.
However, TBM depends on consistent definitions, governed allocation logic, reliable data, and explainable results. AI should strengthen these characteristics rather than introduce uncontrolled variation.
Data preparation and classification
AI can recommend classifications for financial records, identify potential Taxonomy mappings, and help practitioners investigate unmapped or inconsistently classified data.
Recommendations should be evaluated against approved rules and source evidence before they become part of the governed TBM Model.
Data quality and anomaly detection
AI-assisted analysis may help identify unusual spending patterns, missing records, unexpected changes in consumption, inconsistent mappings, and other conditions requiring investigation.
Anomalies are indications for review, not proof of error. Legitimate business changes can produce unusual patterns, and model recommendations must be validated.
Analysis, forecasting, and scenario development
AI can assist practitioners in exploring trends, preparing analytical explanations, developing forecast assumptions, and comparing scenarios.
These applications are most useful when they operate on governed information and distinguish observed facts from estimates, assumptions, and generated interpretations.
Reporting and stakeholder engagement
AI may help translate complex TBM information into accessible explanations, draft recurring reports, and support natural-language exploration of approved datasets.
This can make TBM information easier to use, particularly for stakeholders who do not regularly work with detailed financial or modeling reports.
Any generated interpretation should preserve traceability to the underlying evidence and remain subject to appropriate review.
Suggested callout: AI-assisted TBM requires governed information
AI can recommend classifications, identify patterns, and prepare analyses. It should not independently change approved mapping rules, allocation methods, financial results, or management decisions without appropriate authorization and validation.
Reliable TBM remains the foundation for useful AI-assisted analysis.
Practical TBM and AI Use Cases
The following use cases illustrate how organizations can apply TBM to AI management and how AI can support the operation of a TBM practice.
Each blueprint identifies the management problem, required information, minimum modeling or data requirements, recommended methods, expected outputs, and the decisions it can support.
The first three use cases apply TBM to AI investments and operations. The final three apply AI within the TBM practice. Their technical requirements differ accordingly.
Using TBM to Manage AI
AI Cost Transparency and Allocation
Management problem
AI spending is frequently fragmented across cloud platforms, model providers, software contracts, internal labor, and shared enterprise services. Financial records may identify who purchased a resource without establishing which AI capability consumes it or which business consumer benefits.
This limits budget accountability, demand management, forecasting, and the ability to distinguish productive consumption from avoidable expense.
Decisions supported
- Which AI services and consumers are responsible for spending?
- What drives changes in AI consumption and cost?
- Where should costs be attributed directly or allocated?
- Which spending requires investigation or improved accountability?
Required data
Minimum inputs include general ledger or accounts payable records, cloud and AI-provider billing, usage quantities where applicable, accounting periods, and identifiers connecting resources to applications, services, or owners.
Additional useful information includes SaaS entitlements and assignments, service catalogs, deployment identifiers, cost center ownership, shared-platform consumption, and contract commitments.
Minimum TBM Model depth
Resource Towers provide a useful starting point for identifying and grouping AI-related technology costs. Mapping to Technology Solutions is needed to establish meaningful service-level costs. Technology Consumers are needed when the objective includes allocating costs to business recipients.
The appropriate minimum therefore depends on the reporting objective. A Resource Tower view can support infrastructure transparency, but it cannot independently establish complete AI service or consumer costs.
Allocation methods
Directly attributable charges should be assigned using reliable resource, subscription, application, or deployment identifiers.
Shared inference and compute costs may be allocated using measured workload consumption, adjusted where necessary for meaningful differences in resource demand. Software licenses may be attributed through assignments or contractual entitlements. Shared labor and platform costs may require documented effort, service consumption, or other reasonable drivers.
Where measurement is unavailable, estimates should remain distinguishable from observed usage. Allocated spending must reconcile to the relevant source totals.
Expected outputs and measures
Useful outputs include AI spending by provider, Resource Tower, Technology Solution, and consumer; consumption and cost trends; budget variance; unallocated spending; and the proportion of costs attributed through direct measurement versus allocation estimates.
Relevant TBM Outcomes
Transparency, Insights, Alignment, and Optimization.
Organizational Value Drivers
Primarily Financial Performance and Operational Efficiency. Other Value Drivers may be influenced by decisions made using the information.
Implementation considerations
Begin with a manageable set of AI services and establish ownership and cost reconciliation before introducing complex allocations. Validate resource identifiers, avoid double-counting shared charges, and distinguish embedded AI functionality from separately identifiable AI spending.
Showback can be introduced before chargeback to test allocation credibility and improve stakeholder understanding.
AI Unit Economics and Total Cost of Ownership
Management problem
Organizations may compare AI alternatives using model prices, token rates, or infrastructure costs without accounting for the full resources required to deliver a functioning business capability.
This can distort build-versus-buy decisions, obscure ongoing operating commitments, and make it difficult to determine whether an AI service is becoming more economical as consumption grows.
Decisions supported
- What does it cost to deliver an AI-enabled service or business transaction?
- How do different delivery architectures compare?
- Which components contribute most to total cost?
- Are changes in consumption improving or worsening unit economics?
- Should the organization build, buy, redesign, or replace a capability?
Required data
Required information includes technology costs, usage or output volumes, service ownership, and the relationships connecting underlying resources to the AI capability.
For a fuller TCO analysis, include development and integration labor, software and model-provider charges, infrastructure, data services, evaluation, security, monitoring, operational support, contracts, and expected lifecycle changes.
Business output measures should be defined with the service owner. Depending on the application, these may include completed transactions, processed documents, resolved interactions, or successful workflows.
Minimum TBM Model depth
Resource Towers can support infrastructure-level unit costs, such as cost per GPU hour, when cost and consumption boundaries align.
Technology Solutions are the recommended minimum for evaluating complete AI product or service economics. Mapping to Technology Consumers and business capabilities improves attribution and supports evaluation of business-level outcomes.
Allocation and calculation methods
Assign directly attributable costs to the relevant solution. Allocate shared infrastructure and services using measured consumption where possible, or documented drivers reflecting resource demand.
Calculate unit cost by dividing the relevant total cost by a consistently defined unit of output over the same period.
Technical consumption measures should be normalized only when the underlying workloads are genuinely comparable. Token counts cannot be converted into a universal number of standardized inferences, and differences in model quality or task complexity can invalidate simple benchmark comparisons.
TCO should distinguish one-time implementation costs from recurring operating costs and make the analysis period explicit.
Expected outputs and measures
Outputs may include cost per completed business transaction, cost per AI-enabled service interaction, total lifecycle cost, fixed versus variable cost, cost by delivery alternative, and forecast unit costs at different demand levels.
Comparisons should include service quality, reliability, and relevant operating requirements alongside cost.
Relevant TBM Outcomes
Transparency, Insights, Benchmarking, Strategy, and Optimization.
Organizational Value Drivers
Financial Performance and Operational Efficiency, with potential implications for Experience, Innovation, and other drivers depending on the capability.
Implementation considerations
Define the unit of output before building the calculation. Avoid comparing unlike workloads, confusing infrastructure cost with complete service cost, or treating expected productivity improvements as realized financial savings.
Use sensitivity analysis to evaluate changes in consumption, pricing, service requirements, and implementation assumptions.
AI Portfolio Governance and Value Realization
Management problem
AI portfolios often contain initiatives at different stages of development, with inconsistent business cases, fragmented ownership, and limited evidence of realized benefits.
Without continuing evaluation, organizations may scale initiatives that have not demonstrated sufficient value, maintain redundant capabilities, or overlook opportunities to redirect resources.
Decisions supported
- Which AI initiatives should be funded, expanded, redesigned, consolidated, or discontinued?
- Are expected benefits being realized?
- How do initiatives compare against organizational priorities?
- What financial and operational commitments accompany each investment?
- Where could resources create greater value if reallocated?
Required data
Minimum inputs include an AI initiative inventory, ownership, investment and operating costs, lifecycle status, intended outcomes, and approved business cases or investment assumptions.
Additional information may include adoption and usage, service performance, business KPIs, forecast costs, contract commitments, operational dependencies, risk assessments, and realized benefits.
Measures should reflect the purpose of each initiative rather than imposing a universal AI ROI calculation.
Minimum TBM Model depth
Technology Solutions provide a useful minimum for connecting AI capabilities to their costs and operating responsibilities.
Mapping to Technology Consumers and business capabilities is recommended for understanding who benefits and how investments relate to enterprise priorities.
Some early-stage initiatives may initially be managed through project or investment records before a stable production Solution exists. Those records should remain connected to the relevant financial information.
Allocation and evaluation methods
Direct project and service costs should be attributed to the relevant initiative. Shared platform, labor, and support costs should be allocated using documented consumption or effort drivers where appropriate.
Evaluation should compare expected and actual costs, adoption, operational performance, and business outcomes. Financial benefits should distinguish avoided expenditure, realized cost reduction, capacity creation, revenue contribution, and other forms of value.
A portfolio review should also consider transition costs, contractual obligations, dependencies, and opportunity costs before recommending rationalization.
Expected outputs and measures
Useful outputs include an AI investment portfolio, lifecycle status, spending against plan, forecast commitments, benefit realization, adoption and performance indicators, and investment decision recommendations.
Each initiative should have an identified owner, intended outcome, relevant evidence, and a review cadence appropriate to its maturity.
Relevant TBM Outcomes
Strategy, Alignment, Insights, Transparency, and Optimization.
Organizational Value Drivers
Potentially all six Organizational Value Drivers, depending on the initiative. The relevant drivers should be selected according to the intended organizational consequences rather than assigned automatically.
Implementation considerations
Establish common governance requirements without assuming every initiative requires the same metrics or approval thresholds.
Separate evidence of technical success from evidence of organizational value. Revisit business cases as initiatives mature, and provide explicit decision options for continuation, expansion, redesign, consolidation, and retirement.
Portfolio governance should support learning and resource reallocation, not simply compliance with an approval process.
Using AI to Improve TBM
AI-Assisted Data Classification and Taxonomy Mapping
Management problem
TBM practitioners frequently spend substantial time reviewing financial records, resolving inconsistent descriptions, maintaining classification rules, and mapping source data to the TBM Taxonomy.
These activities can become difficult to scale when organizations have many source systems, changing vendors, inconsistent account structures, or large volumes of unmapped records.
AI can assist by identifying likely classifications and presenting evidence for practitioner review.
Decisions and activities supported
Which source records are likely to belong to particular Cost Pools or Resource Towers?
Which records appear inconsistent with approved mapping rules?
Where are new vendors, descriptions, or accounting patterns creating mapping exceptions?
Which classification recommendations can be approved, and which require further investigation?
Required data
Inputs include source financial and operational records, approved Taxonomy definitions, existing mapping tables, historical classifications, account and vendor reference data, and documented exceptions.
Useful supporting information includes service catalogs, organizational ownership, contract descriptions, and prior practitioner decisions.
The system should retain source identifiers and the classification rules or evidence used to generate each recommendation.
Minimum TBM Model depth
No completed downstream Taxonomy mapping is required to begin. This use case can operate on source records before they enter the TBM Model.
The required target depth depends on the classification task. Initial applications may focus on Cost Pools or Resource Towers. More advanced recommendations involving Technology Solutions or Consumers require reliable relationship and ownership information.
AI processing approach
Use approved rules and existing mappings as the primary reference. AI can recommend classifications for unresolved records, identify similarities with previously classified items, and highlight potential inconsistencies.
Recommendations should include supporting evidence and a confidence or review-priority indicator. Confidence values should not be treated as reliable probabilities unless calibrated and validated.
Approved classifications can then be incorporated into governed mapping tables through established change-control procedures.
Expected outputs and measures
Outputs include proposed classifications, exception queues, records requiring additional evidence, mapping consistency reports, and approved updates.
Useful measures include classification accuracy on a representative reviewed sample, percentage of records requiring manual investigation, exception resolution time, and the proportion of spending remaining unmapped.
Relevant TBM Outcomes
Transparency, Insights, and Optimization.
Organizational Value Drivers
Potentially Operational Efficiency and Financial Performance through reduced manual effort and improved decision information. These benefits must be measured rather than assumed.
Implementation considerations
Begin with a limited classification problem and an approved reference dataset. Evaluate recommendations against known correct classifications before expanding use.
Protect sensitive financial information, retain a complete review history, and prevent AI-generated classifications from silently changing production mapping logic.
AI-Assisted Analysis and Forecasting
Management problem
TBM practitioners must interpret large volumes of financial and operational information, explain variances, identify emerging trends, and develop forecasts or scenarios for management decisions.
These activities can require extensive manual analysis, particularly when changes involve multiple cost drivers, services, vendors, and business consumers.
AI can assist with investigation, interpretation, and scenario preparation when supported by governed data and reliable analytical methods.
Decisions and activities supported
What explains significant changes in technology costs or consumption?
Which patterns warrant investigation?
How might spending change under different demand or pricing assumptions?
What are the likely consequences of alternative sourcing, capacity, or investment decisions?
Which findings should be brought to management attention?
Required data
Inputs may include historical actuals, budgets, forecasts, operational consumption, service and consumer mappings, allocation rules, vendor pricing, contractual commitments, and known business drivers.
Additional information may include project schedules, planned capacity changes, business demand forecasts, and documented management assumptions.
Data should retain time periods, organizational boundaries, units of measure, and source provenance.
Minimum TBM Model depth
No universal minimum applies. The necessary depth depends on the analysis.
Cost Pools and Resource Towers may support infrastructure cost forecasting. Technology Solutions are needed for service-level analysis. Technology Consumers improve demand attribution and business-level scenario evaluation.
The analytical model should use the deepest mapping necessary to answer the management question reliably, without implying that greater detail always produces better forecasts.
AI processing approach
AI can help identify unusual patterns, formulate investigative questions, summarize validated analytical findings, and assist practitioners in constructing scenarios.
Calculations should use controlled analytical methods. Forecasting models should be selected and tested against historical results, with assumptions and limitations documented.
Generated explanations should distinguish observed relationships from inferred causes. AI should not invent drivers to explain an unexplained variance.
Expected outputs and measures
Outputs may include variance explanations, exception reports, forecast scenarios, assumptions registers, sensitivity analyses, and management summaries.
Measures include forecast accuracy, completeness of variance explanations, time required for analysis, and the percentage of generated findings supported by verifiable evidence.
Relevant TBM Outcomes
Insights, Benchmarking, Strategy, Alignment, and Optimization, depending on the application.
Organizational Value Drivers
Potentially Financial Performance and Operational Efficiency, with additional drivers determined by the decisions supported.
Implementation considerations
Begin with a bounded analytical workflow and compare AI-assisted results with established practitioner methods.
Validate calculations independently, prevent unauthorized access to sensitive information, and maintain clear separation between source facts, model estimates, scenario assumptions, and recommendations.
Forecasts should remain subject to accountable human review before they influence budgets or investment decisions.
AI-Assisted Reporting and Stakeholder Engagement
Management problem
TBM information can be difficult for stakeholders to interpret when it is presented through complex dashboards, detailed financial reports, or unfamiliar modeling terminology.
Practitioners may also spend considerable time preparing recurring explanations, responding to similar questions, and tailoring information for different audiences.
AI can help translate governed TBM information into accessible narratives and support more responsive stakeholder engagement.
Decisions and activities supported
What changed in technology spending, consumption, or service economics?
Which information is relevant to a particular stakeholder?
How should significant findings be explained?
What follow-up questions should be investigated?
How can recurring reporting be prepared more efficiently?
Required data
Inputs include approved TBM reports, metric definitions, governed datasets, business glossaries, reporting periods, organizational ownership, and access permissions.
Supporting information may include prior approved reports, known business events, service descriptions, and documented explanations of significant variances.
The system must distinguish information the requesting stakeholder is authorized to access.
Minimum TBM Model depth
No additional Taxonomy mapping is inherently required beyond that supporting the underlying report or question.
Cost Pools and Resource Towers may be sufficient for financial or infrastructure reporting. Technology Solutions and Consumers are needed when explaining service economics, business demand, or consumer accountability.
AI processing approach
AI may draft narratives from approved reports, summarize changes, explain defined metrics, and help users navigate available information through natural-language questions.
Where supported, responses should retrieve information from governed sources and provide references to the relevant data, reporting period, and calculation definitions.
Generated explanations should not introduce unsupported causal claims or present recommendations as approved management decisions.
Expected outputs and measures
Outputs may include recurring management reports, executive summaries, stakeholder-specific explanations, question-and-answer responses, and follow-up investigation requests.
Useful measures include report preparation time, factual accuracy, source traceability, stakeholder adoption, and the proportion of responses requiring correction.
Relevant TBM Outcomes
Transparency, Insights, Alignment, and Strategy.
Organizational Value Drivers
Potentially Operational Efficiency and Experience, with indirect implications for other Value Drivers when better information supports subsequent decisions.
Implementation considerations
Start with a defined set of approved reports and common stakeholder questions.
Apply role-based access controls, require source references for material claims, and establish review requirements for financial or executive communications.
The objective is to make TBM information easier to understand and use without compromising its accuracy, confidentiality, or governance.
Getting Started with TBM and AI
Organizations can begin applying TBM to AI without waiting for a comprehensive AI cost model or a fully developed AI governance program. Similarly, practitioners can begin evaluating AI-assisted TBM workflows without introducing AI across the entire practice.
The most effective starting point is a specific management problem with a clear owner, available information, and an identifiable decision or improvement objective.
Getting started: Using TBM to manage AI
Identify the AI services and investments that matter. Begin with a defined portfolio, service, provider, or business capability where spending, consumption, or value is not adequately understood.
Establish financial and operational visibility. Identify relevant costs, usage measures, ownership, and available service relationships. Determine which information is directly measured and which requires estimation.
Connect resources to solutions and consumers. Extend the TBM Model to the depth needed for the management question, using documented allocation methods where direct attribution is unavailable.
Define the decisions and measures. Establish what leaders need to decide and what evidence will support those decisions. Select unit costs, performance measures, and business outcomes appropriate to the capability.
Incorporate the information into governance. Use the resulting understanding to guide funding, demand management, optimization, scaling, and rationalization decisions. Review results as conditions change.
Getting started: Using AI to improve TBM
Select a bounded practitioner workflow. Choose a recurring activity such as classification review, variance analysis, or report preparation where the expected benefit can be evaluated.
Establish a governed information source. Provide approved data, definitions, mapping rules, and examples of correct outputs.
Define AI’s permitted role. Determine whether AI will recommend, classify, summarize, analyze, or draft—and which activities require human authorization.
Test accuracy and usefulness. Compare AI-assisted outputs with established methods, evaluate errors, and document limitations before expanding use.
Integrate with practice operations. Establish ownership, review procedures, access controls, performance measures, and a process for correcting or discontinuing ineffective applications.
Suggested callout: Begin with a management question
Whether using TBM to manage AI or AI to improve TBM, start by identifying the decision or process that needs improvement. Then determine the data, methods, controls, and evidence required to produce a useful result.
AI, TBM Outcomes, and Organizational Value
TBM Outcomes and Organizational Value Drivers play different roles in managing AI.
TBM Outcomes—including Transparency, Insights, Benchmarking, Strategy, Alignment, and Optimization—provide capabilities that help leaders understand technology and make informed decisions.
Organizational Value Drivers describe the dimensions of value those decisions and subsequent actions can influence.
AI initiatives may contribute to Financial Performance through improved economics, Operational Efficiency through better execution, Experience through improved interactions, Innovation through new capabilities, or other dimensions of organizational value. They may also introduce risks, resource dependencies, or competing consequences that must be evaluated.
The relationship is not automatic. Better AI cost transparency does not itself reduce spending, and an AI-assisted forecast does not itself improve financial performance. Value depends on how leaders use the information, which decisions they make, and what results those decisions produce.
Explore the Organizational Value Drivers and the TBM Framework for a broader understanding of these relationships.
Explore TBM and AI Resources
The following Council publications provide additional guidance on AI economics, technology modeling, investment governance, and organizational value.
Understanding AI economics
Beyond Tokenomics: TBM Delivers the Full Economic Picture for Accurate AI ROI
Explore why technical consumption measures alone are insufficient for evaluating AI economics and how TBM supports a more complete understanding of AI investments.
Managing the AI portfolio
When AI Doesn’t Deliver: Using TBM to Guide AI Rationalization
Examine how TBM can support decisions to improve, consolidate, replace, or discontinue AI capabilities when actual results do not justify continued investment.
Understanding technology resource costs
Technology Resource Towers in TBM Modeling & Technology Value Management
Learn how Resource Towers contribute to technology cost modeling and why connecting resources to solutions and consumers is important for understanding business value.
Strengthening technology investment governance
Transforming Tech Governance into Business Decision-Making
Explore how TBM can connect technology governance with the information and decisions needed to manage investments against business priorities.
Building executive confidence
How the Best TBM Practices Build a Trust Premium
Consider how reliable information, consistent management practices, and stakeholder confidence contribute to the effectiveness of TBM.
Connecting technology to enterprise priorities
Explore how technology management can be connected to the objectives and priorities of enterprise leadership.
Developing the TBM discipline
Building the TBM Body of Practice for the Age of AI
Explore the importance of developing governed, repeatable TBM methods as AI introduces new opportunities and management requirements.
Managing AI as Part of Technology Business Management
AI is becoming part of the technology portfolio and the broader operating environment of the enterprise. Managing it effectively requires more than tracking spending or selecting promising technologies. Organizations need to understand AI’s complete economics, its consumption and operational dependencies, the business capabilities it supports, and the organizational consequences of investment decisions.
TBM provides a common management discipline for developing that understanding and connecting it to governance, resource allocation, and ongoing evaluation.
At the same time, AI offers opportunities to improve the effectiveness of the TBM practice. Applied to governed information and well-defined workflows, AI can help practitioners manage data, develop insights, evaluate alternatives, and communicate findings more effectively.
The opportunity is therefore reciprocal: TBM can make AI investments more understandable and manageable, while AI can help make TBM more capable and responsive. Both contribute when reliable information leads to better decisions and those decisions produce meaningful organizational value.
The principle of designing value from the beginning is critically important to AI solutions which require early, high-stakes decisions on architecture, infrastructure, training approaches, and operating models. Applying TBM only after AI is deployed limits its ability to influence strategy or justify spend. With TBM by Design, AI capacity, TCO, and ROI modeling become inputs to planning and architecture decisions—guiding innovation toward the highest-value outcomes.
How TBM by Design supports AI:
- Incorporates TCO and capacity forecasts into AI funding and business case approvals
- Supports build-vs-buy decisions based on projected costs, performance, and ROI
- Enables early identification of high-cost or low-yield solutions
- Aligns scarce resources (GPUs, data science time, etc.) to priority use-cases
Embedded TBM delivers:
- AI investment governance based on modeled cost and value tradeoffs
- Faster time-to-value with more confident go/no-go decisions
- Greater agility in adjusting AI roadmaps as real costs and results emerge
Getting Value Quickly from TBM for AI
AI doesn’t wait—and neither should your TBM practice. You don’t need perfect models to get started:
- Begin with SaaS-based AI solutions and map costs to basic Taxonomy structures
- Use directionally correct models to identify major cost drivers for more complex AI
- Leverage existing billing, infrastructure, and labor data—even if partial—to create early insights
In many cases, a rough TCO model delivered in the first 60–90 days is more valuable than a “perfect” model 12 months later—especially if it prevents waste or strengthens the case for further investment.
Take the Next Step
Innovation thrives on clarity, alignment, and momentum. TBM brings all three to AI initiatives. Whether you’re piloting chatbots or scaling agentic platforms, TBM gives you the tools to model costs, forecast needs, and prove value.
- Start where you are—model even partial data
- Engage with AI leaders through the TBM Council community
- Embed TBM into the foundations of every AI initiative
Explore more in the TBM for AI Community or dive into the TBM Taxonomy 5.0 to begin mapping AI to value.
