Here’s How I Advise Anxious C-Suites To Approach Generative AI

As a strategic account executive at Salesforce, I spend my days advising companies on how to use emerging technology to grow their business. I can tell you this: C-suite leaders are nervous about generative AI in business. They know they need to do something but don’t have clarity on what, or how. I’ll tell you what I tell them. 

But first, here’s a sampling of what they’ve told me: 

• A chief technology officer said, “We’re well aware of the importance of generative AI, but we’re struggling to figure out how to make it work for our business.” 
• A chief digital officer said their teams brainstormed close to 100 high-level use cases, posing a challenge in prioritization and execution. 
• A chief information officer said he’d make exceptions to the company’s standard budget process to fund any generative AI technology with a one-year payback.

There’s no mistaking it: Generative AI is the top priority, and every organization I’ve spoken to has elevated it to a boardroom-level discussion. It’s natural to feel discomfort and anxiety around a technology that’s fast-moving and changing rapidly. I know that when I’m overwhelmed with a huge decision, I break it down into steps — and that’s what I’ve done for the C-suite leaders I advise. Here are six steps to follow as you navigate generative AI.

Step 1: Identify the right use cases for generative AI in business

Any application of generative AI has to start with the question, “what business problem am I trying to solve?” 

To find a relevant use case, review your current processes from end to end, and identify any friction points — those processes that lead to frustration, wasted time, lost opportunities, and in some cases, staff turnover. A tip: Focus on the problem, and avoid the temptation to fit the solution to the problem. As alluring as generative AI is, it may not be the right solution for every problem.

Here’s an example: As part of their cost-saving measures, a client of mine asked their IT leader to lower call center expenses without adversely affecting customer satisfaction. Their customers were experiencing long wait times (a key cause of customer attrition) while call center representatives tried to find answers to their questions. Initially, the company wanted to use generative AI to instantly scour knowledge articles for the right answer; however, after some analysis, they realized that predictive AI was the better fit, allowing them to arm agents with insights such as a customer’s propensity to churn.

As you evaluate your current process, think of generative AI as supercharging your workforce, powering it to work faster, and more efficiently and creatively.

Step 2: Prioritize for impact, then execute and iterate

Once you’ve identified one or more use cases for generative AI in business, prioritize them by considering these factors: 

• ease of implementation
• strategic importance
• potential revenue
• cost savings
• time savings
• overall impact
• customer and employee satisfaction

Which of these factors is most important to you? The answer will be different in every case. For one of my clients, a chief medical officer of a large neurological institute, the patient journey is at the center of their generative AI use case. Their initial focus is the post-operative care division, which receives highly technical calls from patients that the staff is not generally trained to answer. This knowledge gap affects patient care. 

The first iteration of the solution for this use case is a chatbot to address high-level questions from patients. Future iterations will evolve the chatbot’s responses with contextual knowledge and more nuanced patient questions. The company’s vision is to transform the post-operative care division within two to three years, enhancing its staff’s capabilities and equipping them with more advanced tools to best serve patients.

The most successful organizations are embracing curiosity and learning as they adopt generative AI in their business. They focus on impact and set regular checkpoints to optimize for value. They adopt a beginner’s mindset and thrive on an iterative approach with short development cycles and consistent feedback.

Step 3: Create a playbook

If you’re struggling on the implementation of the use cases, an organizational playbook can help. This playbook is your comprehensive guide. It outlines your organization’s approach to rapid evaluation, iterative testing, effective cost management, impact measurement, goal setting, continuous improvement, security, and expansion into various generative AI applications. 

The playbook should include specific guidance like how generative AI fits with other applications the team uses, as well as general guiding principles that give your teams flexibility to push boundaries and get creative. Additionally, consider establishing a Center of Excellence (CoE) that fosters collaboration between business and IT teams for the creation and review of this playbook. Business teams might own business problems the applications solve for, while IT might own infrastructure set up and security. The teams would join efforts around continuous improvement of the applications and output quality with regular audits, testing, and touchpoints. 

Implementing generative AI in your business can be overwhelming, so give people time  and space to focus on learning, adapting, and creating, and allow them to fail and grow.

Step 4: Develop a strong data strategy

A solid data strategy is key to the success of AI in business. Ultimately, accessible, high-quality, secured data is essential for high-quality output. If your data is in any way flawed, the information the AI gives you will be, too. 

When fine-tuning AI models, make sure the AI has easy access to relevant data. That means the data is accurate, updated, and complete. 

The CIO of a leading media company told me he was excited about having a conversation with their data by asking it questions to inform strategy. 

Chief data officers face the challenge of helping companies get the most value from generative AI. A strong data strategy ensures high-quality data throughout the lifecycle: building enhanced capabilities into the data architecture; securing sensitive data while staying compliant with evolving regulations; and investing in data engineering talent. 

Salesforce Data Cloud, for example, provides the data for Einstein Copilot, Salesforce’s new generative AI conversational assistant, and makes sure the outputs are contextually relevant. Using Data Cloud and Einstein AI, you can create targeted marketing segments, personalize website landing pages based on consumer behavior, and customize emails for specific campaigns with the help of Einstein Copilot.

Step 5: Determine what success looks like

You can’t know if you’re successful if you haven’t defined what success is. That’s why I advise companies to determine this at the outset. 

A senior vice president of a large technology company said his company’s biggest threats are a) not adopting new ways of using AI fast enough and therefore falling behind, and b) finding and retaining talent with AI expertise. 

They’re exploring whether AI can use data analytics to improve their business. For example, they hope AI can help them quickly process market, competition, and customer data so they can optimize products and services more efficiently. In this pursuit, they’re defining the parameters of success and pinpointing key performance indicators (KPIs) to assess the technology’s impact. 

As you build your generative AI use cases, define your metrics and categorize them as primary and secondary KPIs. The primary KPIs are your business KPIs. For example, your service department might look at minutes saved on customer calls. Sales and marketing departments might look at sales qualified leads and growth in marketing, respectively.

Did generative AI help your company meaningfully affect these KPIs? That’s what you’re looking for. Do you have higher-quality leads and are you converting them at higher rates? 

The secondary KPIs are your generative AI KPIs. These include accuracy/error rate, output quality, training time, scalability, training cost/resources, and productivity gains.

If you don’t hit your targets, go back to the beginning and reassess your use case. Was generative AI the right solution for the problem? If so, consider whether you used the right foundation model for the task, whether you trained it on the right data, whether that data was high quality, and whether it was grounded in business context relevant to the task, among others. 

Step 6: Bring your business and IT teams together

If your company hasn’t aligned its business and IT teams yet, now’s the time. Doing so will let your teams focus on well-defined priority use cases for generative AI in business, and will make sure solutions are tailored to meet tactical and specific business objectives. Earlier, I mentioned establishing a CoE to write your playbook, and in general, a CoE is a great direction to take: It can be your collaborative hub for shaping best practices and ensuring that generative AI aligns with your company’s strategic goals.

It’s abundantly clear generative AI is not the flavor of the month — it’s the future we’re stepping into. We know you’ve got questions and concerns. With the above guidelines, think about how you want to reshape your current processes and tap into AI’s power.