Impact of AI on Business Value

With the arrival of GenAI (including Agentic AI), no organization will remain the same. For many, fundamental changes are already underway. GenAI is both an opportunity and a threat.
With GenAI, results can be achieved in a way that wasn’t possible before, but it also brings risks of a type and scale we have never seen before.
So, how should organizations apply GenAI to truly improve business value and increase the delivery speed? It starts with a business vision and strategy, which lie beyond the scope of quality engineering. The next step is to focus on activities that contribute to achieving the desired business value. This is where quality engineering, and specifically the application of quality measures, comes into play.

Expectations

Business managers expect GenAI (including Agentic AI) to lower costs and shorten timelines. Some have (too) high expectations, while others think this new hype will pass.
The expectations relate to innovation, value and speed.
In this new era, organizations will spend much more money on information technology, but not through traditional IT channels. GenAI promises to directly support businesspeople without involvement of IT professionals. GenAI also brings opportunities that never existed before. To benefit from those promises and opportunities, business management and IT management need to align closely to get the anticipated innovation, value and speed.

As with any other new technology, managers should realize that having new technology available does not mean that this technology will automatically solve their problems. Sometimes, not using the new technology is the best decision. However, in the case of GenAI, simply ignoring it does not seem like a smart decision. GenAI is here to stay, and organizations that decide not to use it will quickly lose their relevance.
Choosing how to apply GenAI in a responsible way is one of the tasks that quality engineering can support. For example, by supplying the perspective of business value and the related quality risks, and by pointing out the limitations around the use of AI, such as legal limitations of using AI in high-risk situations.

Getting a Grip on the Complex New Situation

Top-level management feels the pressure. GenAI promises incredible results but also comes with risks that are difficult to fathom. So, it’s all about balancing business value and quality risks.

What problem are we solving?
The core question is: “What problem are we solving?”
Once this is clear, the next question becomes: “Can GenAI be (part of) the
solution?”
And finally: “How can we deal with the associated risks?”

Can GenAI be (part of) the solution?
The problem to be solved must be linked to the pursued business value. An organization has a mission and vision that provide focus for its activities.
The result of the organization’s activities should generate some form of business value. This could be monetary value, or it could take other forms, such as improving the sustainability of products or services.

How can we address the associated risks?
The introduction of GenAI-based systems in an organization increases the complexity of the IT-systems. This requires careful consideration. Also, the introduction of GenAI may increase the complexity of business processes.
However, when well applied, GenAI may also simplify business processes. People in the organization will need to get used to these changes and the resulting complexity must be well-managed.
GenAI tools generally perform tasks much faster than humans would, which promises to increase the speed of IT delivery.

Changes in the complexity of business processes and IT systems are some examples of quality risks. If you are involved in introducing GenAI in your organization, pay attention to careful management of those risks. Do keep in mind that in some situations, the best solution is not to apply GenAI. Traditionally coded solutions may be more reliable and faster in situations where the problem can be solved with an algorithm.

Don’t Become a Fool With a Tool

One of the big fears when introducing GenAI is that people stop thinking for themselves and blindly accept all results GenAI produces. To cope with this phenomenon, organizations need to actively equip their people to stay in control and avoid blindly delegating tasks to tools. Remember the old saying “a fool with a tool is still a fool,” which can be extended especially with GenAI in mind to “a fool with a powerful tool is a dangerous fool!!”
GenAI is a tool that can empower people when they consciously make choices about when to apply it and when not to. The requirement of human oversight when GenAI is used, as mentioned in the EU AI Act, is based on the principle of knowledgeable experts being involved in such decision-making (we call this “human expert in the lead”).

The VOICE Model Provides Support

TMAP’s VOICE model is a clear way to take the pursued business value as starting point for all activities in the IT delivery process.
Based on this model, various steps in IT delivery are organized to support the decision-making process of the various management groups involved. This VOICE model is applicable in any situation, including when GenAI is part of the IT solution. On the one hand, this model helps maintain an overview and minimize the process’s complexity. On the other hand, it provides clear
starting points for all kinds of IT delivery and (especially) quality engineering activities.

Value: Any IT system aims to bring value to someone. This value must be defined; this often reveals implicit expectations and makes them explicit. Keep in mind that high quality alone is not necessarily high value. Quality means different things to different (groups of) people. Sometimes, there is more value in a system with a lower quality level that is quickly available than in a high-quality system that is available too late.
For some people the mere fact that GenAI is part of the solution already is business value. Most people, however, would want to see specific added value of the introduction of GenAI.
Objectives: To understand the purpose of an IT system and how to create and maintain it, quantifiable objectives with an adequate level of detail must be established.
Indicators: Whether the objectives are met, and whether the pursued value can be achieved, needs to be measured. To this end, indicators are specified that will be measured by testing and other quality measuring activities. GenAI may support in measuring the indicators and amplify the use of feedback.
Confidence: The results of measuring the indicators will provide essential information for stakeholders to gain confidence that the IT system will be able to achieve the pursued business value.
Experience: Once the IT system has been incorporated into the operational business processes, the users will experience the actual business value. Based on this, they may require further improvements, and a new cycle of the VOICE model, also known as the “value-improvement loop”, is triggered.

When implementing new IT systems, or when adapting existing IT systems, for example by adding GenAI-based functionality, an incremental approach will often offer significant benefits. By starting small, the first business value can be achieved quickly. And by having regular retrospectives and evaluations, the solution is gradually extended, which increases the business value. This is achieved with the value-improvement-loop of the VOICE model of TMAP.

Employees May Be Disappointed by Dedicated GenAI Solutions

Because of significant risks associated with introducing generic GenAI tools (such as ChatGPT or Gemini) or specialized GenAI tools (such as coding tools like Windsurf or GitHub Copilot) in an organization, many organizations limit the possibilities of the tools. For example, they choose to implement a dedicated GenAI tool that has specific guardrails for secure use, such as limiting the access of the tool to the internet (so that any data that is entered will not leak outside the organization) or by limiting the tool’s functionality.

In some organizations, internal versions of generative AI tools are developed to meet requirements around security, compliance, or data residency. For example, a company might build its own solution using the OpenAI API within a secure Azure environment, often referred to as a “Wrapper GPT”. This approach offers more control but also comes with trade-offs.
These internal tools are usually tailored to support specific business tasks, but they may lack the broader capabilities found in public tools like ChatGPT. Internal teams may not only move slower than providers like OpenAI, but they are also often limited by internal policies, budget, or technical constraints. As a result, advanced features such as multimodal input (working with images or video) might be unavailable due to these limitations or compliance concerns.
Some organizations even completely prohibit the use of GenAI tools.

In addition, prompting the model to explore diverse reasoning paths can be beneficial for uncovering edge cases and alternative perspectives. By converging these different lines of reasoning, the model is more likely to produce a well-rounded and comprehensive answer.

These limitations may cause negative feelings among employees who are used to having full access to GenAI tools on their private devices. If users are accustomed to the full public version of, for example, ChatGPT, they may seek workarounds using personal devices or external tools to access the features they miss. This introduces business risks. Not only is it unclear who is using which tools and when, but it becomes impossible to control what data is shared with external systems something that may directly conflict with your organization’s data security policies.
To maintain employee support, it is important to explain the reasons for limiting the possibilities of GenAI tools and provide guidelines for alternative uses, such as using GenAI tools on personal devices for performing business tasks.