Originally published in Toward Data Science.

I get asked by executives and entrepreneurs what they should be doing about artificial intelligence (AI). I give talks on the topic and I think I have the start of a framework for executives and entrepreneurs alike on how to find near-term AI opportunities. The AI Sense and Respond Framework distills the current things that AI can do with into actionable business things an executive or entrepreneur can do.

The framework breaks down AI capabilities into three sets of boxes (for simplicity):

  1. Pattern Recognition
  2. Prediction
  3. Automation

Note: the third item (Automation) doesn’t have to involve AI algorithms per-se, but it is simply the actions that can be taken based on the results of the first two (pattern recognition and prediction).

Sense: Pattern Recognition & Prediction

Pattern recognition and prediction are categorized in a section I call Sense. Under the Sense section, we have your data inputs, your trained models, and outputs that can drive insights and decisions. The input data could come from your current datasets, other companies’ datasets, sensors, etc. The data you have available is based on your own business and will be different for everyone. The goal of the Sense section (pattern recognition and prediction) is to perceive something about the world that can be interpreted by the AI models. Pattern recognition (including classification) and prediction are reasonable, applied things that AI-related models can do now. The models take inputs and perceive something good enough to lead to insights and decisions.

Respond: Automation

The other section is called Automation. It is in a section called Respond. These are the (optional) actions that the computer can take (automation) on its own based on the findings (insights and decisions) from the pattern recognition and prediction data that was generated from Pattern Recognition and Prediction.

Using the Framework

To use the framework, you will want to list the 3 sections across the x-axis and categories of your business along the y-axis. I used a very simple example of two general categories that all businesses have: Generate Revenue and Operational Efficiency (you can use Marketing, Sales, Accounting, Operations, Sales, etc. — whatever works for you.)

Now, fill in each Pattern Recognition and Prediction box. These are things that applied AI can do now given the right information. Now, once these boxes are filled in — you can end there — and you will have useful use cases for AI in order to generate insights that can lead to decision making. You would think about what patterns you would look for to generate more money (for example), and so on. These are use cases where AI technology can be used to benefit your business. Then, you can complete the Automation box as well. Read through the Pattern Recognition and Prediction use cases and ask “what action could we take next”. This will help you come up with things that can be automated based on AI-driven insights.

There are other complexities around implementation, of course. You might find out that you need to clean up old data or collect new data in order to create the software that you want. But, this framework will help you think through what you can do and how it can help you.

Building Sense and Respond Systems

Ultimately, we’re building a sense and respond world. I may write more about that in a future post. Essentially whether we’re talking about IoT (the Internet of Things) or any business, we’re talking about about building learning systems. Inputs with pattern recognition, predictions, observations, and outputs. And, then going back and doing it all again in a never-ending learning cycle. The output is understood in context with observation (Sense) and the cycle starts again. All while allow for human input/insights while providing its own insights. All of our companies are learning systems. Some will just do a much better job of it in the future. AI techniques can help all businesses because it fits well into the cases where the system needs to understand patterns or make predictions. Which, are both fundamental keys to having a system that actively learns.

I think this framework is useful for big enterprises and could generate a lot of internal AI-driven projects that can bring value to your company. Also, entrepreneurs can use this model to generate new AI startups to meet the needs of customers by matching needs you have observed from customers with capabilities that AI can provide. So, best of luck building your own AI Sense and Respond based use cases.