Harnessing the Power of AI: Converting Hype into Business Value

The past three years AI has taken a massive leap in popularity, pushed by the amazing achievements of companies like OpenAI, nvidia and Google DeepMind. Especially the low barrier of entry for OpenAI’s ChatGPT has propelled the world into the age of generative AI.

It takes nothing more than visiting a website, and then you can interact with some of the most advanced AI tools ever created, be it for content, imagery or advise. Which is how many people use it.

Converting AI to value is hard

As reported by resumebuilder that 49% of all companies in US are using Chat GPT - an additonal 30% are looking to use it in 2024. That’s a mind numbing amount of companies who are using one single tool. This is the hype-train in full motion. But are they converting ChatGPT/AI to business value? Sure, there’s the immediate benefits for an employee

• Project Planning Assistance: Guidance on planning and organizing projects.

• Presentation Guidance: Helping with creating or improving presentations.

• Text Rewriting Help: Assistance in rewriting or optimizing content.

• Coding Support: Support for coding tasks and debugging.

All these are apparent and instant wins for the employee. If you’re not using a generative AI like this, you will fall behind in race for jobs. Simple as that.

But is this really large scale business value? No.

Identifying the problem

Most organizations fail to identify the problem they want to solve. They start at “We want to use AI” and then they try to invent some use-case. This is poor strategic planning at best, and a waste of company money and employee times at worst. You may end up finding an appropriate case to which you can apply AI for a solution, but more likely than not, you will probably find a mediocre problem which is not worth solving.

How do you identify the problem to be solved? Like any other business problem.

Find pain points

• Identify and interview stakeholders - what is an immediate pain?

• Do a process or value chain analysis - Do we have bottlenecks or miscommunications? Where? Why

Scope the problem

• Create a clear problem statement/pain point

Our customer support response times are too slow, with an average first response time of over 24 hours, resulting in a 15% customer churn rate.

• Determine which business units are suffering from this problem - is it worth solving? In the case above the immediate verticals that come to mind are:

  • Support
    They are probably working inefficiently or the process can be optimized. They are most likely overworked because of the massive backlog of tickets. Reducing the workload and making it easier for the support teams to be a success will definitely increase team morale and employee satisfaction while also increasing customer satisfaction.

  • CSM
    The CSM team try their best to upsell, but if there’s a feeling that you don’t get help with what you already struggle with, why would you purchase more? They are already behind on points in every conversation they have.

  • Sales
    Word of mouth is amazing, as long as it’s not because your product is trash. Take PostNord in Denmark for instance, they delivery a fairly robust service and 90% of all packages are delivered within a day - yet they struggle with an image as being absolute garbage for the last years. No wonder it’s difficult to get any business.

  • Development/Product
    The development department probably has to chip in more than once to assist the support or make changes to help them answer the plethora of tickets. Probably the solutions are more so information sharing than a lack of product-fit. Releasing overhead in cross-organizational communications frees up resources everywhere.

• Make a decision - is it worth solving or not?

Feasibility/fit analysis for AI

Many leaders with no understanding of AI will say something along the line of;

Can’t we just take hook out data up with ChatGPT and then it can spit out the solution?

This is ignorant and overlooks the fundamental requirement for any machine learning system: structured data. Before leveraging AI, you need to ensure that your data is organized and governed properly. For instance, if your ERP system contains duplicate items/goods, you'll need to clean that data to prevent contamination.

If you the foundation is actually in place, you can begin considering the next challenges:

• Expertise Availability: Determine if your team has the necessary AI expertise.

• Infrastructure: Check if the current infrastructure supports AI development.

• AI Solutions Identification: Identify potential AI approaches (e.g., machine learning, NLP, computer vision).

• Alternative Solutions: Consider non-AI solutions as benchmarks.

• Compliance: Are you properly understanding the compliance and privacy concerns with AI

All of which is at least as complicated as simply identifying if you actually have a problem that AI can help with or not.

Conclusion: Converting Hype to Value

Congratulations, you've identified a problem without boarding the AI hype train. You're on the path to executing an AI project that creates real value, not just buzzwords. However, remember that AI isn't always the solution. Without structured data and clear governance, AI projects can quickly fall short of expectations.

Instead of starting with "We want to use AI," focus on identifying clear pain points and evaluating the feasibility of AI as a solution. Only then can you unlock the true potential of AI in delivering measurable business value.

At the end of the day - the low barrier of entry for generative AI, tricks people into thinking it’s a low barrier of entry to solve business problems with it as well. That is not the case - there’s many challenges and even more unique to various use cases and flavors of AI solutions, all of which we haven’t even touched here. But if you find your self with a senior stakeholder saying “We need to use AI because it’s amazing!” then you can drag them through this process before indulging in their ambitions and ignorance.