AI Strategy & Business Case

AI Management: A Checklist for Executive Management

Marin Niehues5 November 202612 min read

Marin Niehues in a striped shirt, thoughtfully resting his hand on his chin.

Executive management steers AI effectively once it has answered eight questions with certainty, from the objective through risk and law to the metric in regular operation. The checklist arranges these questions along the path an AI initiative takes from a delivery perspective: understand, build, introduce. Each question can be answered with yes or no. Each no names a task that should be complete before the next investment decision.

The eight test questions at a glance

The checklist consists of eight questions that you answer with yes or no in an executive management meeting, assigning an owner and a deadline to every no.

A survey by the German Federal Statistical Office shows that in 2025, 26 percent of companies with 10 or more employees already used AI, rising to 57 percent among companies with 250 or more employees. In many companies AI emerges from below, through individual tests, licenses and strategy papers. Executive management steers only when these parts converge into one portfolio with a sequence, a budget and one accountable person. The checklist tests this connection.

It is the test instrument for the principles covered in the article AI management: what executive management has to steer. That article sets out the steering tasks of executive management. This one shows how to tell whether those tasks are being fulfilled. Because question 7 touches on legal matters: this article provides an overview and does not constitute legal advice.

Understand: where AI creates value

The first question concerns the target picture. It is usable if it can be put into one sentence that names a business value, such as shorter lead times in order management or lower handling costs in service. The second question concerns the portfolio. A business analysis collects the candidates and clarifies their requirements. It then assesses each candidate by value and feasibility, with risk as a third axis. The result is a sequence. Applications with high value and little dependence on other systems come first, because their results justify the funding of the next stages. The article Prioritizing AI use cases describes the method in detail.

The third question concerns the business case. It calculates the expected benefit in euros or in hours saved and sets it against the full costs. These include licenses and operation as well as the enablement of users. Where the benefit requires a change in behavior, the business case names that change and the point at which it will be measured. A business case without an assumption about the usage rate overstates the benefit, because it counts every user as an active user.

Build: from prototype to operation

The fourth question tests whether a prototype demonstrates the benefit in the actual workflow. A prototype with real users on real transactions produces measurements even when its scope remains small. The fifth question concerns operation. The transfer into operation succeeds with internal or external teams, provided three points are settled in advance:

  • The operation of the solution has a named owner.
  • The maintenance of the data foundation is assigned to a team.
  • Requests for change follow a fixed decision path.

Where these points stay open, the pilot phase extends without a decision on regular operation.

Three legal questions belong on the checklist: the classification of every application under the European AI Act, the protection of personal data and the AI literacy of staff.

The AI Act classifies AI systems by risk. Prohibited practices are excluded, high-risk systems such as those used in recruitment or credit scoring are subject to strict requirements, and transparency obligations apply to systems that interact with people or generate content. A company that uses such systems counts as a deployer. For high-risk systems, the regulation requires deployers, among other things, to ensure human oversight by competent persons and to use the system in line with the provider's instructions. Executive management therefore needs a register listing every application with its classification. The article EU AI Act gives an overview of the regulation.

Irrespective of risk class, Article 4 of the AI Act obliges providers and deployers to take measures ensuring that their staff have a sufficient level of AI literacy. The regulation prescribes no particular training format. What the obligation requires in detail is described in the article AI literacy under Article 4 of the AI Act.

The General Data Protection Regulation applies to every processing of personal data, including by language models. For each application, the legal basis and the data processing agreement with the provider must be settled, and for higher risks a data protection impact assessment. For executive management itself, the general duty of care is what counts: it must be able to show that risks have been identified and that responsibilities and controls are in place. A maintained register with classification as well as owner and review date is the simplest evidence.

Introduce: accountability and metrics

The sixth question concerns accountability. For each application, a named person in the business unit, often as product owner, answers for the result. A delivery manager steers the initiative across departmental boundaries and keeps decision paths short. An introduction is complete when the workflow has changed, with adapted process descriptions and role-specific enablement of users.

The eighth question concerns metrics. Two levels suffice. Usage metrics show how many users apply the solution regularly. Impact metrics show what changes in the outcome, for example lead time or handling cost per transaction. Only impact justifies further funding.

A fixed rhythm connects the eight questions. Executive management reassesses the portfolio every quarter: which applications are running, which assumptions from the business case have materialized, which sequence applies now? How governance and budget interact with the AI organization beyond the individual initiative is set out in the pillar guide AI management.

How does executive management use the checklist as a template?

The checklist becomes a template when you transfer the eight questions into a table with columns for yes or no, owner, deadline and evidence, and place it in the papers for the quarterly meeting.

Three steps suffice for the first use. First, each member of executive management answers the eight questions alone. The meeting then compares the answers, because differing assessments reveal where clarity is lacking. Finally, every no receives an owner and a deadline ahead of the next investment decision. The evaluation follows a fixed rule. A no on questions 1 to 3 halts new investment until it is resolved. A no on questions 4 to 6 halts the roll-out of the application concerned. A no on question 7 halts use until the classification is available. A no on question 8 permits continued operation but requires measurement before the next meeting.

Niehues Consulting as a partner for AI management

Niehues Consulting supports executive management on precisely this undertaking from a delivery perspective. Founder Marin Niehues works as a management consultant, builds organizations that deliver and brings AI into operation. His projects range from VC-funded scale-ups to DAX 40 groups, with assignments in 20+ countries on six continents and a home base in the DACH region.

The approach follows the Niehues Delivery Model with its phases Understand, Build and Introduce, which also structure the checklist. For executive management, three of the model's review questions matter most. Which business outcome should change measurably? Has the business case been confirmed with real values? Which metric shows after 90 days that the solution is landing?

On request, Niehues Consulting steers the initiative itself as delivery manager, project manager or product owner.

The entry point is a free initial consultation of 30 minutes, in which the eight test questions are applied to your company. You can arrange it via contact.

Frequently asked questions

What does AI management in a company include?

AI management includes business analysis and the business case, the selection and sequence of applications, and the steering of the portfolio with metrics. Classification and control of risks are also part of it. The aim is a portfolio that executive management leads through decisions.

Who is accountable for AI within executive management?

Overall accountability rests with one member of executive management, who represents the sequence and budget of the portfolio. For each application, a product owner in the business unit answers for the result. These two levels keep decision paths short.

How often should executive management review the AI portfolio?

Executive management reviews the AI portfolio quarterly, because the assumptions of a business case can first be checked against usage data within a quarter. Initiatives with a high budget additionally receive a short monthly review of their metrics.

When is an AI initiative ready for regular operation?

An AI initiative is ready when the prototype has met its acceptance criteria, defined in advance, in the actual workflow and an owner has been named for operation. The metrics are running, so that the impact is visible from the first day. The classification under the AI Act is documented.

What are executive management's obligations under the AI Act?

Executive management ensures that every AI application is classified by risk and that the obligations of the relevant class are met. The AI literacy obligation applies to every company that uses AI. For high-risk systems, human oversight and use in line with the provider's instructions are among the additional duties.

Is AI training for employees mandatory?

Article 4 of the AI Act obliges companies that use AI to ensure a sufficient level of AI literacy among their staff. The regulation prescribes no particular training format. Role-based training with documentation is the usual way to meet the obligation verifiably.

Does a company need an AI policy?

A company whose employees use AI tools needs an AI policy. It sets out which tools are approved, which data may be entered and who reviews results. The policy also serves as evidence for the literacy obligation.

Does a company need a Chief AI Officer?

A dedicated Chief AI Officer makes sense once the portfolio spans several divisions and has its own budget. Below that threshold, a named member of executive management with a clear mandate for the AI portfolio suffices. What matters is one person with authority over budget and priorities.

Is the checklist available as a PDF or template?

The checklist in this article is built as a template and can be adopted directly. Transfer the eight questions into a table with columns for yes or no, owner, deadline and evidence. In the free initial consultation with Niehues Consulting, it can be applied directly to your portfolio.

Part of the guide: AI management: the guide for executive and divisional leadership

Let’s talk about your initiative.

In 30 minutes we clarify where your organization stands and which step will pay off most now.

Request a consultation

30 minutes · free · no obligation