Strategic management of AI use cases

Tobias Bohnhoff
Mar 1 · 7 min read

Artificial intelligence (AI) is an often-cited buzzword, but what does the technology actually mean for day-to-day business? In a nutshell, AI is the combination of three elements: A valid and clearly defined problem or question, suitable data sets and an algorithmic model that translates the question into processable analysis or calculation steps. The result is a data-supported suggested solution with corresponding probabilities about the reliability and generalizability of the statement.

AI is not limited to the use of modern neural networks, which have to be operated on extremely powerful processors with large amounts of data, but also includes optimization methods and classification algorithms, which have been part of the working environment for years. Directly connected is the area of Business Intelligence, with the important difference that the translation into decisions takes place more directly, i.e. either fully automatically or near real-time.

Key requirements for generating actual value from AI:
1. AI is not solely an IT issue, it needs top-management support
2. Companies need a data strategy
3. Organizational flexibility, implementation processes never run linearly

The intelligence is often not located in the algorithms, but results from business-relevant and precise questions as well as suitable and high-quality data sets. The algorithm is merely the translation into execution. However, this requires a high level of expertise on the part of the responsible data scientists or data engineers to perform this translation between a strategic business perspective and mathematical processing.

If you look at AI use cases from a structural point of view, you can define different subsets of enterprise functions in which AI can be applied to help boost your company’s performance. In our research at Appanion Labs, we identified 46 different use cases throughout 8 major enterprise functions. Of course, this is not a comprehensive list and you can find various other niche use cases. But it gives a framework perspective and almost every use case that we came across so far can be mapped to one of the categories displayed in this overview:

Requirements to achieve value from AI

To benefit from artificial intelligence means to question existing processes without anxiety. This is particularly difficult for companies where current processes are well-established, or individuals see their role threatened by change. Data-driven operations do not mean working without people, but rather enabling collaboration between people and software or machines. In concrete terms, this means that the effective use of AI should take three core aspects into account:

First, AI is not solely an IT issue. It requires top-level management support to adapt processes and minimize uncertainties in the enterprise. Furthermore, evaluation methods for determining development and test budgets must be adapted. Often the focus is one-dimensional on new business or cost reduction. However, using AI means transforming processes in order to benefit from it in the long term. Of course, this also includes the first two aspects, but additionally factors such as improved product quality, minimized risk of failure, higher employee satisfaction and reaction speed or scalability of processes.

Secondly, companies need a data strategy. An essential result of this is the identification of where in the company the use of artificial intelligence is particularly worthwhile. First of all, it must be recorded how the previous added value is distributed across different corporate functions such as marketing, sales or production in the organization and how this structure will change in the future as a result of external influences or the company’s own strategy. In a second step, a complementary “map” must be created for the existing data sets in the respective company functions. The volume, the data quality and the contextual linkage of these databases have to be considered. Using AI where a lot of data is available, but hardly any added value takes place, makes as little sense in the first step as attempting to initiate projects in high value-added areas without having suitable databases.

Third, implementation processes never run linearly. They require iterative and close cooperation between the specialist department and technical experts. Prototypes for the validation of suitable data resources and suitable algorithms can be the key to successful application development. Cross-functional teams, fast intermediate results and organizational appreciation to perceive potential break-off points in a project as learning success, are necessary organizational characteristics in this context. This also promotes a modern working environment in order to attract the best talents for the company.

Manage AI projects according to an innovation/purpose matrix

When looking at AI use cases the key question remains, which horse to back in order to derive success and ROI. AI can be used very broadly as an element of incremental process optimization in the best performing corporate divisions as well as technological enabler for the next growth segment of a company. Depending on this role, the evaluation of budgets and focus of development areas changes radically. The following concept is based on Geoffrey Moore’s performance zone concept:

Incubation

Most companies start to evaluate in incubation mode, meaning that a lot of ideas are collected and prioritized to start small experiments or build prototypes. Afterwards, the results are categorized into ideas to kill, ideas to watch, ideas to scale and ideas to integrate in the existing business as performance boosters. This field of incubation is clearly a cost center and should focus its investments on ambitious future-oriented projects, but also be able to easily shut down projects, if they fail or are not suitable for further procedure.

AI use case focus: Broad scope, testing should focus on ideas that have the potential to scale to become a relevant new business segment. In this context, projects can also start industry and use case agnostic but should aim to get proof-of-concept in one focus area.

Management recommendation: Commit an incubation budget, select the most promising topics to explore and make it easy for project owners to get funding based on the target to present a proof-of-concept in order to decide how to proceed.

Transformation

The transformation zone is the rocket launcher, the place for disruptive innovation and moonshot projects. Focus and commitment are important in this area. Transformation zone is not about experimenting, but only about scaling a proven concept to the point where it is “moving the needle” for a large company. This point is usually reached when the new business line contributes roughly 10% or more to the overall revenue. By that time it should also be able to grow easily at double digit rates in order to be considered as “next big thing” internally — but of course also for external stakeholders, the stock market etc…

AI use case focus: Normally focused on the industry, if diversification is not clearly defined management priority. Use case focus has to be build out through the transformation zone, not suitable are use cases without direct revenue impact, e.g. HR, risk management, controlling etc.

Management recommendation: This is a one-shot opportunity, if a topic is pushed into transformation zone it should get all the support of the organization that is required, in terms of human and financial resources. It cannot be built up incrementally, paused or run on priority two — it has to be 100% commitment or shut down. This requires a profound technical feasibility check and a strong business case.

Productivity

The productivity field might be one of the most interesting fields for AI. It comprises all enterprise functions not directly generating revenue such as risk management, marketing, supply chain and logistics etc. but it offers at the same time a lot of potential for process optimization and automation enabled through AI. Prioritized by indicators such as the current process costs, number of employees involved or criticality to fulfill the overall value-add of the company’s product or service. Spendings in AI development should be structured accordingly to the expectable ROI.

AI use case focus: Industry focus is less important in this context, the priority is on enterprise functions without direct revenue impact such as HR, security management, supply chain optimization, accounting etc. Successful projects have the potential to build out own licensing services to competitors or other verticals, which then would be a case for transformation zone.

Management recommendation: The Challenge here is to not only focus on cost-saving issues through the automation of processes, but also to encounter better decision-making, higher product quality, risk reduction or higher process flexibility. Ideally, projects were already tested in incubation zone.

Performance

Performance zones are the cash cow segments of the business. The areas where 80 to 90 percent of the company’s revenue are allocated. To change processes within these segments is highly critical because of the impact caused on the overall business performance, stock price development, valuation etc. Therefore, the goal in this segment is to use technological innovation researched and tested in the incubation zone to apply to existing process in order to incrementally improve efficiency, effectiveness or compliance of those processes.

AI use case focus: Strong focus on industry and one dedicated use case that impacts the revenue generation directly, e.g. in sales, product marketing, cash flow optimization etc. Initiatives should be reduced to one project by the time to filter out the most promising one and avoid organizational confusion.

Management recommendation: Not a space for experiments, use cases applied to the performance zone have to be fully proven in previous R&D steps and need to work reliably in terms of the used algorithms, data quality and computing infrastructure. Ideally, no-harm testing takes place with at a smaller scope before the application is scaled.


For further information on strategic AI use case development feel free to contact us at appanion.com.

Tobias Bohnhoff

Written by

Founder at appanion.com. Technology enthusiast and passionate about trends and innovation in artificial intelligence.

APPANION

APPANION

At Appanion, we develop strategies, prototypes and tangible action plans to work successfully with emerging key technologies such as artificial intelligence.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade