Scaling the AI Use Case Hunt

C-Levels must understand: Finding viable AI use cases isn’t just a Google search away.

#beyondAI – I'm deep in the AI bubble, and I like it. AI is my comfort zone. I’ve been involved since 2010, and professionally since 2014. It’s second nature for me to spot problems that AI can solve and to quickly grasp new technologies in the field.

But that’s just me.

In most companies, 99% of employees aren't comfortable with AI yet—whether it’s the technology itself or recognizing potential business opportunities for AI. And I think that’s only fair. Companies are made up of diverse specialists. Even generalists, like AI Product Managers, are specialized generalists in their own way.

Despite this, many of my friends and peers in the field report that their companies are launching major automation and augmentation initiatives. The common challenge? Scaling the process of identifying viable AI use cases.

I completely get this struggle.

When you have a few AI teams already working on projects, who should be responsible for spotting additional AI opportunities across the organization?

And I’m talking about real AI opportunities, not just AI use cases. You can easily Google or ask ChatGPT for use cases in your industry and get a bunch of ideas.

Today, I’m sharing my perspective on this challenge and why I believe that scaling the AI use case hunt isn’t the real issue.

Happy reading! 🛋️

P.S. If you’ve found my posts valuable, consider supporting my work. While I’m not accepting payments right now, you can help by sharing, liking, and commenting here my LinkedIn posts. This helps me reach more people on this journey, and your feedback is invaluable for improving the content. Thank you for being part of this community ❤️.

---

---

The Paradox of Scaling the AI Use Case Hunt

Typically, companies rely on one of three approaches:

• Hiring consultants to find more opportunities.

• Adding this task to already overburdened teams.

• Asking business areas to identify automation potential.

None of these approaches are ideal. At least not in isolation.

Consultants often bring in use cases that worked for other companies, but what succeeds elsewhere doesn’t always succeed in another organization. They probably even tell the hiring company this. But the allure of AI use cases is hard to resist. I’ve written an entire article on this—“The AI Use Case Pandemic”—and I think it’ll help you see where I’m coming from.

Overloading your existing teams with the task of identifying AI opportunities is equally problematic. These teams are often already stretched thin, focused on delivering current projects. Adding more responsibilities can lead to burnout and stifle creativity and innovation. Imagine an AI Product team setup like this:

• Data Scientists, Data Engineers, AI Engineers, and an AI Product Manager are building a company-wide Churn Prediction Model. This is a full-time job for each of those specialists. How can they actively scout for new AI opportunities? At best, they might stumble upon new ones while building their own product—but that’s more accidental and passive than intentional.

Then there’s the option of asking business areas to come up with AI opportunities themselves. I understand this approach—starting from the problem and finding a solution is crucial. And who better than domain experts to articulate their problems? But there are two challenges here: One is the same as above—overloading teams with extra tasks doesn’t work long-term. The other is that business teams often lack the understanding of what’s automatable or augmentable with AI. Even IT experts not directly working with AI struggle to identify the right AI cases within their own processes.

The Bureaucratic Catch-22 in AI Assessments

Companies also need to recognize that problem discovery and business opportunity evaluation can be time-consuming, regardless of the technology used to create a solution. You have to streamline and adapt your processes to enable teams to conduct these explorations. Don’t force teams that are here to help your organization become better to fit into processes that aren’t designed for AI development.

There’s a sweet irony in the corporate world: Your AI teams need approval for a business case to fund detailed assessments of an AI use case, but they can’t create that business case until the detailed assessments are done. It’s like needing a loan to prove you’re creditworthy but needing creditworthiness to get the loan. 🤦‍♂️

This is the kind of corporate logic that keeps things interesting, right? 😅

Cheers to all those stuck in this bureaucratic loop—just another day in the life of "agile" enterprise decision-making!

💡 What Do We Mean by Detailed Assessment? When we talk about "detailed assessment" in the context of identifying viable AI use cases, we’re referring to the in-depth analysis required to evaluate whether an AI idea is worth pursuing. This involves looking at multiple factors: the technical feasibility, the availability and quality of data, the potential business impact, and the resources needed for implementation. It’s not just about saying, "This could be done with AI," but about rigorously assessing whether doing so would be effective, efficient, and aligned with the company’s goals.

This process often requires cross-functional collaboration, drawing on the expertise of data scientists, AI engineers, business analysts, and domain experts to ensure that the AI solution is not only technically possible but also valuable and sustainable in the long run.

Yes, this paradox is surprisingly common in large enterprises. The requirement to have detailed assessments before being able to create a business case, while simultaneously needing an approved business case to fund those assessments, is a well-known bureaucratic catch-22. It often stems from rigid processes and risk-averse cultures, where decision-making is heavily dependent on thorough analysis and multi-layered approvals.

This situation is particularly prevalent in organizations with complex structures, where different departments and stakeholders must coordinate closely but are often hampered by procedural bottlenecks. It highlights the challenges of balancing the need for thorough evaluation with the need for speed and agility in decision-making.

But this was just one example, here are a few others I have collected over the last 10 years:

• Data Access Dilemma:

  • Situation: Your AI team needs access to company data to build and train a machine learning model, but the data security policy requires that only approved AI models can access sensitive data.

  • Catch-22: You can't get the data to build the model without approval, but you can't get approval without demonstrating the model’s effectiveness using that data.

• Budget Approval Loop:

  • Situation: An AI project requires a budget for initial research and prototyping, but the finance department won’t release funds until there’s a proven ROI.

  • Catch-22: You need to run the prototype to demonstrate ROI, but you can’t start without the budget, creating a loop that stalls the project.

• Pilot Project Paradox:

  • Situation: The AI team is required to show the impact of AI on a small-scale pilot project before getting approval for larger implementation, but leadership insists on seeing results that can only be achieved at scale.

  • Catch-22: You need a full-scale deployment to demonstrate impact, but you can’t scale up without first proving the concept on a small scale.

• Vendor Selection Stalemate:

  • Situation: Your AI initiative requires specialized third-party tools or services, but the procurement department won’t approve the purchase until you’ve demonstrated that the tools are essential and effective.

  • Catch-22: You can’t prove the tools’ effectiveness without using them in a live environment, but you can’t acquire them without prior approval.

Many companies recognize this issue and are working to streamline their processes, but it remains a common pain point in the corporate world.

---

Empowerment: The Key to Unlocking AI’s Potential

What’s needed is a more integrated approach—one that leverages the unique insights of your employees while fostering a culture of innovation throughout the organization. This could involve training programs to raise AI literacy across departments, encouraging cross-functional collaboration, or creating dedicated roles for AI opportunity scouting within teams.

By empowering employees at all levels to understand and identify AI use cases, companies can discover opportunities that truly align with their specific needs and objectives, rather than relying on generic solutions or overburdened teams.

And this is where AI Product Managers can play a crucial role. They excel at identifying business opportunities within problems, blending business acumen with AI expertise. But they usually focus on one AI product at a time, once they find the right opportunity. Also, the role of an AI Product Manager isn’t as widespread or prominent in companies as Data Scientists or other AI development roles.

Building a coalition of experts—like AI Product Managers, Data Scientist Leads, and other experts—could be a solution. Together, they could create a platform where domain experts can seek advice on their problems.

Another idea is creating a company-wide AI Use Case & AI Solution inventory. Think of it as an actively managed portfolio. If you see an AI solution built for marketing to tackle churn in that portfolio, you might consider scaling it to the service department, where agents can see if a customer calling in is prone to churn.

But creating such platforms and inventories isn’t a decision that AI Product Managers or other experts can make alone. It’s up to the decision-makers—CEOs and executive board members—who are eager to transform their organizations into ones that achieve operational efficiency. They need to understand that finding viable AI use cases isn’t just a Google search away. It requires a systematic, product- and user-centric approach. It demands collaboration between domain experts and AI experts and sufficient space to explore potential opportunities. Concepts like those mentioned above might be the solutions for your struggles and ambitions. But you need to know that you can’t delegate this decision away. You must empower your employees to perform in the way you envision.

If finding business opportunities—whether for cost reduction or revenue growth—were easy and risk-free, everyone would have already achieved operational excellence.

---

So dear senior executives, help your experts to help you.

The challenge isn't in scaling AI use case identification—we tackle far more complex problems every day. What might seem like rocket science to many is just daily business for us.

The real issue lies in empowerment.

Fix that, and trust your experts to handle the rest.

JBK 🕊

---

P.S. If you’ve found my posts valuable, consider supporting my work. While I’m not accepting payments right now, you can help by sharing, liking, and commenting here my LinkedIn posts. This helps me reach more people on this journey, and your feedback is invaluable for improving the content. Thank you for being part of this community ❤️.

---