The Solstice AI Cookbook: The Cognitive-Ready Meal

It’s a cook-off. You versus a master chef, the likes of Grant Achatz, Stephanie Izard, or Rick Bayless from our hometown of Chicago.

You’re given the same final dish description. You’re provided the same budget for ingredients but are responsible for finding them on your own and limited to the tools available in your respective kitchens.

Think you stand a chance? For the vast majority of us, the answer is no. Although a world-class dish may fit nicely on a single plate, it’s the culmination of a deep understanding of flavor, years of training, access to top-tier tools, and even the courage to challenge convention.

The artificial intelligence (AI) race is in full swing, and businesses all have their eyes on the final main course: agile business models with greater efficiencies and dynamic new customer experiences. But knowing what you have a taste for isn’t the problem — it’s knowing how to create the magic yourself: how to perfectly balance a set of ingredients using the most advanced tools and expertise to create something truly remarkable.

With that in mind, we’ve put together the Solstice AI Cookbook, because as any top chef will tell you, the key to a great dish lies in the difference between following a recipe and understanding it.




Quality Ingredients


Let's start with ingredients. If the objective of your cook-off is an eggplant parmesan, the ingredients might be listed as eggplant, eggs, bread crumbs, tomato sauce, and basil. But where a master chef has an advantage is in
access to the best ingredients. AI operates in the same way. The raw data is what makes it all possible, and if your data is low-quality or stale, no AI magic can save the day. But that’s only half of the ingredient equation. Deep subject matter expertise and experience is why masters and amateurs are starting from two completely different places.

Here’s a look at some of the most staple types of data, ones that any business must make good use of in the age of digital transformation:


Quality Ingredients



Subject Matter Expertise

 

It’s easy to dismiss or overlook the data your company already has because it might feel like it’s neither special nor interesting enough. But just like in the kitchen, a great chef knows how to use every ounce of their supply. Business forms, for example, are often forgotten about and left to rot. In reality, many of these forms contain free-text boxes that can capture a significant amount of unstructured insight. These fields usually relate to an important business interaction such as a purchase, sales call, or customer query and will manifest as comments, notes, or other information.

 


 

The Tools of the Trade

Any high-end kitchen will have an array of cooking tools — from the most elementary of knives and cutting boards to cutting-edge dehydrators, immersion circulators, and donabes. The range of utensils, instruments, gadgets, and gizmos can be dizzying. AI is no different. There are enough technologies, platforms, and services to make even the most seasoned experts’ heads spin.

We believe it’s always critical to master the fundamentals first. Once we’ve determined exactly what we’re creating and the ingredients we have available, we then identify which algorithms, providers, and models are right for the job. Here are the tools we love to start with:  

Quality Ingredients

3 things you can do with these:

  • Predict
  • Recommend
  • Detect

As with ingredients, it can be easy to overlook the tools right in front of you. Some of the most extensively used elementary technologies can still provide the most value in some situations.

Just because a spatula is simple doesn’t mean it isn’t the right tool for the job. The hard part is knowing what tools to use and how to consolidate them to create something extraordinary.

Finding the Right Cooks

Anyone who has opened a newspaper in the past year knows there are some legitimate concerns over the future impact of AI on the job market. Whatever the future may hold for AI and your business, we believe people will always be what makes progress possible. Our CXDD (Customer Experience Driven Development) process reflects that belief.

Like any great head chef, business leaders must scour the globe to find the right mix of talent and expertise to help them create brand-new experiences from scratch. With a team aligned to the same vision, strategy, and dedication, it might even surprise you what your business is capable of.

So how do we put it all together? How does the perfect batch of ingredients, combined and enhanced by true experts with world-class tools at their disposal, create something lasting and extraordinary? Well, how about a recipe?

 


 

High-Value Asset Risk Management Recipe

Recipe

We’ll start with a potential outcome in insurance, an industry full of structured and unstructured data as well as mounds of subject expertise on hard-to-spot trends. This is an example that can be easily tweaked to match the ingredients and cooks your business has in the kitchen.

  • Problem
    • Standard inspections of high-value assets (e.g., airplanes) can sometimes uncover compliance issues. When noncompliant assets must be sidelined, they can no longer fulfill their obligations, which is costly for insurers that must pay out policies. Aside from policy expenses, insurers need to pay to fly and accommodate technical experts and adjusters wherever the asset is being kept, further compiling costs.
  • Ingredients
    • Maintenance forms (semi-structured)
    • Subject matter expertise
  • Tools
    • Clustering
    • Cognitive reasoning model
  • Solution
    • By running historical maintenance forms through a clustering algorithm, we are able to identify key themes in the text. Once we uncover these themes, we can consult experts to create a cognitive model that determines a risk score based on cluster themes and structured data from the forms. After we create these scores, we can then set a threshold based on historical data to classify an asset as “at-risk.” We can then set up this process to run in real time, identifying risk as soon as maintenance reports are uploaded.
  • Prep time
    • Six weeks to first proof of concept (POC)
  • Outcome
    • The assets will be sidelined at a lower frequency, driving costs down and adding value to insurance customers.

AI will continue to transform the fundamental ways business is done. That much we know. And although it can be challenging to understand where to begin, the key is to approach it like any recipe: one step at a time.

Looking for a recommendation on where to start? Email Ryan Maguire, Solstice’s Director of Artificial Intelligence at rmaguire@solstice.com. He’ll have just the thing.

This post was co-authored by Karl Hampson.