Unlock AI Potential in your company: 4 steps to identify and launch the first AI project for midsized companies (part 1)

The time is now to embrace AI in your business

Artificial intelligence (AI) is having a dramatic impact on the way organizations operate. Companies are starting to understand that by leveraging AI technologies, they can fight for new competitive advantages where ever they work with data or information.

However, when deciding to implement AI into their organizations, business leaders often struggle to find the right entry point use cases to start with.

Identify the right entry point to succeed in the long-term

At the same time, making the first steps right is often a key deciding factor that forms the attitude of companies towards new technologies and, thus, can be the deciding factor that will dictate the pace and success of adopting AI for the organization in the coming years.

This guide will help to point you towards properly identifying the areas where AI can bring the most value or solve the most pressing problems in your business so that you can get inspired on where and how to start with adopting AI.

We have broken down the process for you into four phases:

1. Draft your AI strategy
2. Gather use-case ideas
3. Prioritize your options
4. Select and launch your first use case

1) Draft your AI strategy

Understand what AI is and what it is not

The first step of the strategic phase is to understand what AI can and cannot do.

A basic breakdown of main AI disciplines looks like this

• Natural language processing
• Computer vision
• Machine learning

The methods in each can be further broken down into tens of categories and often overlap.

Start small so you can achieve big things later

The list of potential uses of each method is limited only by human creativity, and the availability of data and research is moving forward fast.

Some notable uses of AI are, for example, generative models that can create text (such as ChatGPT), images (Dall-e) or Stable diffusion, expert systems such as driving systems for Autonomous vehicles, various types of data labeling and categorization, Social AI that tries to understand humans through a range of their facial or other expressions, Robotics, various applications in healthcare such as AlphaFold that focuses on protein folding and so on.

These are often impactful solutions focusing on large or even humanity-important problems that can just showcase the power of AI.

Target pressing problems or obvious opportunities

For businesses, however, smaller and way simpler implementations can solve pressing problems.

When integrated effectively, AI can increase efficiency and accuracy, freeing up time for more strategic and creative work.

It can also provide valuable insights and predictions based on vast amounts of data, improving decision-making and competitiveness. AI can help companies stay ahead of the curve by automating mundane tasks, analyzing data, and identifying trends.

However, it's important to understand that AI is not a one-size-fits-all solution, and its capabilities are limited by the quality of data it is fed and its programming.

Understand your business and industry well

That implies larger problems at later stages of AI maturity within a company. However, identifying initial high ROI use cases does not pose such a challenge.

Besides understanding what the general industry-agnostic use cases in various business functions such as finance or HR are, it’s a good idea to look at other industries to see how AI is being successfully implemented for their specific needs. Here are a few examples: 

1. Manufacturing: predictive maintenance, quality control, optimizing supply chain operations.
2. Transportation: route optimization, autonomous vehicles, predicting traffic patterns.
3. Energy: demand forecasting, grid optimization, and predictive maintenance of power plants.
4. Healthcare: AI-assisted diagnostics, drug discovery, and personalized treatment plans.
5. Agriculture: precision farming, crop monitoring, weather prediction.
6. Finance: fraud detection, algorithmic trading.
7. Retail: personalized shopping recommendations, demand forecasting, and inventory management.
8. Education: personalized learning, grading, assisting teachers with administrative tasks.

These are just a few examples. AI is being used in many other industries as well, and its applications are constantly expanding. 

Set the strategy

Large companies have a more difficult time setting AI strategies, as this means pulling together all kinds of initiatives, data, information, and teams, which always gets more complex across large organizations. It requires plugging data analytics into strategic areas of companies that usually span across various functions, and it might be difficult to find isolated use cases valuable enough to make a case for proper analysis and implementation.

However, for mid-sized and even smaller companies, strategic activities are more often seen even in simpler (shorter) processes while still displaying high ROI business cases for the application of AI.

However, a good strategy still needs to be based on understanding which area of business produces strategic data and information and where is the most room for improvement in terms of cost structures, inefficiencies, product development, or revenue growth. All this needs to be tied back to current and desired company capabilities (and what should be outsourced), leadership buy-in, IT and analytics systems, ethical, social, or regulatory questions, and security concerns.

Move fast to pilot projects

The spectrum of options is large, so don’t let too thorough research bog you down at the strategic phase and move on quickly to defining your desired outcomes and considering how AI can help you achieve them.

2) Gather use-case ideas

The ideation phase of identifying an AI use-case opportunity is where creativity and strategy come together. It is the stage where the business starts to imagine the possibilities of what AI can do for the organization, working on conceptualizing how to bring those ideas to life. 

The ideation phase can be broken down into a few different idea sources.

Pain Point Driven AI Ideas

The first set of ideas can come from identifying pain points in the organization that ultimately lead to lost revenues or decreased efficiency and negative impacts on profitability.

There are a few key steps that can be taken to identify pain points in the company: 

1. Conduct surveys and interviews: This can be done with employees, customers, or other stakeholders to gather information on the challenges they face. This data can provide valuable insights into where AI solutions could make the biggest impact.
   
   
2. Analyze business data: Companies can also analyze data from various sources, such as sales, customer service, and operations to identify areas where performance is falling short. This data can help the company identify patterns and areas where AI solutions could improve outcomes.
   
   
3. Evaluate current processes: Companies can take a close look at its existing processes and procedures to identify areas where AI solutions could automate tasks or streamline workflows. This can help the company identify where AI solutions could increase efficiency and reduce errors.
   
   
4. Benchmark against competitors: Companies can also compare its performance with its competitors to see where it is lagging behind and where AI solutions could help close the gap.

Identify the key metrics you want to improve. To narrow down your options, involve various teams, such as sales, marketing, operations or customer service in the process. By taking an end-first approach, you can ensure that the AI solution you choose aligns with your organization's goals and maximizes its potential impact.

Try to focus on second-order thinking as well. For example, your company might be struggling to make informed business decisions, which can be explored by using methods like fishbone analysis or 5-whys to get to the root cause of the problem. You might end up finding that the problem is related to manual data entry resulting in poor data quality, which can be solved by AI.

The pain-centric approach also helps position the AI use-case opportunity as a strategic solution that addresses a real business need, building the foundation for developing a strong business case and the buy-in of key stakeholders.

Data Availability and Data Source-Driven AI Ideas

There are many potential AI use-case ideas that can be derived from analyzing data to uncover patterns and trends. Analyzing data is a crucial step in uncovering patterns and trends that can inform an AI use-case opportunity.

Ask “What can we do with our data?” (meta-data analysis)

Meta-data analysis, which can involve looking at what kind of data is being collected, helps identify what kind of model can be built around it so that the data tells you something useful.

Ask “What is our data telling us?” (data analysis)

Another approach is to look at what the data is actually saying. For example, maybe a food delivery company identifies in the data that there are a lot of late deliveries taking place, which could lead to the conclusion that it needs a more efficient AI model for traffic and delivery time prediction. 

A team of domain expert, machine learning specialist, and data engineer is often needed to identify the most relevant data sources for your potential AI use case. The data then needs to be extracted, cleaned, and processed by data scientists to ensure it's in the correct format for analysis. There are many tools on the market that help with data analysis, such as data visualization software or machine learning algorithms that uncover patterns and trends. And these tools can help identify the trends and patterns that help determine the AI use case.

Online AI Use-Case Libraries

Online AI use-case libraries, such as that of Gartner, McKinsey or AImultiple.com, can serve as an excellent source of inspiration when determining your own potential use cases. These libraries provide a wealth of information on AI solutions and applications across different industries and can help business leaders find a potential use case that matches their needs. 

Here are a few steps that business leaders can follow to use these libraries effectively:

• ​​Search by industry: Many AI use-case libraries allow users to search for use cases by industry, such as healthcare, retail, finance, etc. This can be a great starting point for business leaders as they can easily see the AI solutions that are relevant to their specific industry.

• Look for case studies: Case studies are a valuable resource for business leaders as they provide real-world examples of how AI solutions have been implemented in other organizations and the impact they have had.

• Assess the potential impact: Once business leaders have found a potential use case, they can assess the potential impact that the AI solution could have on their organization. You can look at factors such as the size of the problem the solution addresses, the complexity of the solution, and the resources required to implement it.

• Reach out to experts: Many AI use-case libraries provide a way for you to connect with experts in the field. These experts can help you understand the details of the use case, including the technical aspects, and provide guidance on how to implement the solution in your own organization.

By using online AI use-case libraries, you can find inspiration about what companies similar to yours have identified as opportunities and start exploring the potential benefits of that AI solutions within your environment

BONUS: Often overlooked type of “data”: Unstructured information

One approach your company can use to identify AI use-case opportunities is to examine the areas where employees currently work with unstructured information.

This can include both internal information, such as data stored within the company's systems or external information, starting from simple Googling to external databases, repositories, libraries, or other types of resources.

People usually need to process, assemble, evaluate, draw conclusions, or make decisions based on such unstructured information - a perfect use-case for natural language processing and deep learning methods.

Look for activities that require human intuition or judgment in order to be completed. These types of activities can often be time-consuming and repetitive, making them ideal candidates for automation using AI, so-called intelligent automation.

To determine the specific use cases for AI, it is important to take a closer look at the information being used and the processes involved. For example, if a company is using customer service data to identify patterns and trends, AI could be used to automate the process of analyzing the data and presenting the results in a useful and actionable format. Similarly, if employees are spending a significant amount of time reviewing and assessing customer feedback, AI could be used to automate the process of categorizing and prioritizing the feedback for follow-up.

‍

Continue to Part 2 (to be released)