In cooking, the quality of the dish is often connected to the quality of the recipe used and having the right ingredients. Done correctly, a good dish leads people to enjoy their dining experience and not think about all the detail and effort that went into creating the meal. The “backend” work should be invisible to the diner.
In many ways, artificial intelligence (AI) is like cooking. You start with sourcing the right ingredients (data), and the AI manages the mix, timing, and sequence of those ingredients — while always keeping in mind the most important part, the consumer experience. Based on my experience, there are three pieces to the “AI recipe” that delivers an exceptional AI experience: 1) Create advocates, 2) Focus on the critical few, and 3) Manage AI as products.
AI is an exponential technology that can help solve many of today’s complex business problems. AI is prevalent in industries such as retail and entertainment, but adoption within industries that are highly regulated, like healthcare, can be challenging.
For AI to make an impact, you have to run it as a business, starting with knowing who your customers are. While some business leaders are still skeptics regarding the value of AI because they don’t understand it, don’t believe in it, or see it as a threat it’s essential to identify — or even better create — advocates. That means going beyond the AI algorithms and looking for partners that you take along the engagement journey from curiosity, to trust, and then advocacy, because success will only come when the business leader co-owns the solution with the AI team.
As we started our AI journey several years ago, I found a partner in the Medicare business of a leading health plan willing to see how AI could improve the healthcare journey of their health plan members. I needed to be humble about learning the business and the leader’s pain points. I also had to respect the business context and this leader’s expertise so I could ask the right questions to see how AI could address the pain points, which helps me set expectations on what AI could address and what it cannot.
Using AI algorithms, our team produced a data-driven list of members we believed would be more receptive to outreach. This resulted in meaningful impact in helping members adhere to their medication regimen and helped build this Medicare leader’s trust in AI’s capabilities. This trust led to additional successful projects with the leader, who quickly became an advocate for our work and encouraged others to partner with us to solve their business problems.
It’s important to note two things with this Anthem example. One: I deliberately choose an initial project that was not too complex, because this better positioned the AI team for a quick win, which is important when leaders don’t have the resources or patience to wait a long time for results. And two: I picked projects where I also could get line-level buy-in. You need to people on the frontlines who also support your work because they’re the ones who be impacted by any changes.
Focus on the critical few
You may have heard the warning “Don’t try to boil the ocean.” That holds true when using AI to solve business problems. Early on, it’s important to focus AI efforts on specific projects, allowing the AI team to develop high-quality, actionable insights. In my experience, the opportunities are in specific areas:
Prioritization: answering the question “What does the business want to accomplish, and which outcomes are the objectives?”
Personalization: taking AI capabilities and applying them in a way to drive better customer engagement and experiences
Anomaly detection: leveraging the power of AI to detect outliers and identify opportunities that may have been missed otherwise
Automation: combining AI with traditional business logic and business knowledge to improve efficiency
However, before you start working on technology, you must first know the problem you want to solve. When you meet with a client, you start with understanding their problem, not presenting your solution, which may or may not be the answer. As the opportunity becomes clearer, it takes discipline to not overengineer the solution, using complex AI with simple query statements will suffice.
Manage AI as products
Effectively delivering AI to address business needs requires an agile, multidisciplinary team that consists of several specialists at a minimum: a data scientist working on math, a data engineer working on integration in workflows and experiences, a product leader who can articulate the demand and translate that into the right problems and solutions, and a business subject matter expert. This team creates repeatable data assets so solutions can scale in an affordable way and address the areas of a business problem that AI cannot solve, what we call the “first mile” and “last mile” problem.
The “first mile” involves conducting the right upfront assessments, asking the right questions to solve the right problems. When my team worked with the Medicare leader on the member outreach and engagement issue, they were more likely to succeed because we had an associate who was knowledgeable in Medicare, one who was a strong product lead, and another that had the AI know-how.
Once the AI solution has been successfully developed and deployed, it’s important for the AI team and business to collaborate in bringing that solution into the “last mile,” which is integrating the effort into normal business workflows. This is an essential step to deliver value to the business.
As AI professionals continue to work with their business partners to advance their organization’s goals, teams must position AI as a resource to solve business problems. Keeping in mind these three components to the “AI recipe” will make the product more valuable to the business and the business will see AI professionals as their partner. On the other side, business partners should keep the “AI recipe” in mind as they collaborate with AI professionals to improve their businesses. Following the AI recipe will make for a stronger collaboration and a quality product.