5 steps for baking AI into your enterprise

AI is driving digital transformation in business and the wider economy. For organizations adopting digital transformation strategies that leverage AI, lessons from the kitchen can help inspire best practices and help ensure success

sort filter sift flour separate bake
Thinkstock

Increasingly, digital transformation is key to nearly every enterprise strategy—and it’s prompting the C-suite to embrace and apply new technologies, like artificial intelligence (AI), to traditional business processes.

But AI is not a silver bullet. Solutions require a combination of data, technology, infrastructure, talent and vision. And at times it can get complex.

It’s not all that different from baking, another messy composition process where many things need to come together for success. For your first venture into AI adoption, below are five baking-inspired steps to consider—inspired by our own time in the kitchen.

1. Plan your bake

AI isn’t a single, monolithic technology but is instead a diverse set of technology ingredients with the potential to deliver many types of benefits. Think of yourself stepping into the best test kitchen on the planet, equipped with stainless steel appliances, shelves of recipe books and fully stocked pantry and refrigerator. The possibilities are almost endless.

Enterprise AI innovation also offers numerous opportunities for value creation. Petabytes of structured and unstructured data, and new tools emerging every week to extract useful signal from it. But too much choice can be counterproductive—and focus is your friend.

When scoping and planning, try to express your expectations in a single sentence. If you can’t summarize clearly or explain goals to a colleague in less than two minutes, it’s a red flag. Partners with a clearly articulated initial goals are the ones who are most often successful.

If you are not sure where to start, existing KPIs can be a good reference point. For example, trying to reduce average customer service call time by a few percentage points, reduce churn or simply to explore if there is some signal (value) hiding in your company’s data exhaust or on social media. Whatever it is, find a clearly stated, measurable goal.

2. Deploy savvy help

AI takes advantage of the latest technology, but it’s dependent on the people who work with it and train it, and who understand its advantages and limitations.

Before venturing into AI, find out if your company has already adopted it in some capacity. It could be that your company already baked “cupcakes”—an agile term for small innovation artifacts. While your own group could be completely new to AI, it’s possible other parts of your organization are experimenting with it.

In my experience, spending time reaching out to people across the organization is a very good investment. Pick up the phone and call around, or jump on Slack and search for key terms. A few keyword searches can quickly locate communities and conversations of interest.

Innovation spaces and hackathons can be great ways to cut across organizational silos. My company’s own IBM Cloud Garages are spaces designed to stimulate learning and sharing. Here, approaches like pair programming, where two developers work together to design, code and test, facilitate the transfer of knowledge and expertise. With AI, this collaboration is key, just as a sous chef learns from and supports the executive chef. Co-creation of cakes, or code, results in knowledge transfer and team building. 

3. Leverage recipes

Recipes are the embodiment of the hard-won experience of prior cooks—sometimes going back decades or centuries. The finest French pastries and delicious Thai curries owe much deliciousness to many years of testing and tuning. They can provide a great starting point. For the tentative cook—simply follow and execute as instructed. For the more adventurous chef, explore substitutions and new combinations. Experiment.

Design patterns and reference architectures are the recipes for AI. For teams new to the technology, they can help deliver some quick wins to build momentum, and also help stakeholders agree on the services and developer kits that are most aligned to business needs.

As the effort evolves, the teams will want to make substitutions, create their own recipes and become more experimental and innovative.

4. Prepare your ingredients

Whether it’s in a professional kitchen or at home, the concept of mise en place, preparing your ingredients and ensuring everything is in place before you begin, is key to success. When implementing AI, good preparation offers similar benefits.  Nothing can slow down a sprint like finding out the data is bad, or that platform cannot connect to essential services because of IT issues. Tools, talent, and data ready at hand can be a powerful momentum builder and innovation catalyst. Preparation enables innovation.

Let’s use data as an example. Typically, enterprise data must be wrangled, cleaned, and prepared for the best results. Data preparation can often mean dealing with small or big data sets that are unstructured and live in disparate locations throughout your organization and, in some cases, external data.  

If the objective is to validate a specific method, or start to build skills within an organization, there are many great tools that provide ready-built code and pre-scrubbed data (sometimes public domain) to work with, an “Easy Button” for initial innovation. IBM’s Data Science Experience (DSX) is a good resource with examples of this—including articles, notebooks, data sets, and tutorials—ready to execute.

5. Test before serving

Is your cake fully baked? You’ll need to test it. Agile organizations should regularly test and audit processes and procedures to ensure they remain valid, are well-defined, executable, and implemented as designed. You don’t just set it and forget it. Testing is part of the feedback loop—it’s integral to the scientific method and integral to being a good cook in the kitchen. And it’s an essential method of learning.

Agile organizations are not static. They constantly seek opportunities to improve. As the marketplace becomes increasingly competitive, technology like AI won’t be optional, the marketplace will demand it. At Crédit Mutuel, a leading European bank, before deploying IBM Watson technology across its 5,000 branches, the bank first completed an intensive one-year training pilot with 150 client advisors in 20 branches. Crédit Mutuel was clear on its goal to strengthen customer relationships with improved customer service. The test period allowed the bank to refine its AI “recipe” to one that worked across each location. Watson-based solutions are now being used by 20,000 client advisors across Crédit Mutuel operations in France.

AI is driving digital transformation in business and the wider economy. For organizations adopting digital transformation strategies that leverage AI, lessons from the kitchen can help inspire best practices and help ensure success.

This article is published as part of the IDG Contributor Network. Want to Join?