What CEOs Need to Know about AI
My first September article answers a top question from CEOs - how should I think about implementing AI/ML in my company?
BONUS: I host a live chat in Substack every Friday at 9am PST. This is exclusive to my Substack subscribers! Questions? Don’t hesitate to reach out.
It seems like AI is dominating the news cycle lately, with stories about massive investments in a few players, regular announcements about new versions, and dire predictions of AI ascendancy. What are we as CEOs to think about all of this? How should we and our companies proceed? I’m kicking off my September Leadership article series with a look at AI; what it is, why your company should start using it, and how you can get started.
What Is AI?
AI has been around since long before the 1990s. Remember learning about Alan Turing (1912 – 1954), the British mathematician and logician often considered the founder of artificial intelligence and cognitive science. Students of philosophy may strenuously disagree; however, it’s clear that his criterion for whether an artificial computer is thinking has been used as a benchmark for AI ever since. [1] In simplest form, a Turing test boils down to whether one can distinguish a human mind from an artificial mind through a series of increasingly advanced questions. It’s clear that this benchmark is effectively a moving target. However, we can simply define AI as a catch-all term for machines that mimic human intelligence and cognitive functions.
Machine learning (ML) is a subset of AI that very simply makes predictions based on input data. All ML models require a defined set of data, on which the model algorithm (the framework/program underpinning how the predictions are made) is trained. Once trained, subsequent data inputs will produce the most likely output based on the training data. Typically the algorithm stays constant and the model is updated through the addition of new training data over time. The data can be anything, numbers, text, visual elements, etc. and the model output is consistent with the input data. Large Language Models use text content from a broad swath of text sources to produce text content consistent with the input question, which is frequently referred to as a prompt.
Agentic AI is a system of interconnected components that can autonomously perform tasks on behalf of users. [2] At least some of the components are AI models; others may be commonly available tools such as webpages, software macros, or other elements of business and lifestyle process automation. What makes agentic AI different is the ability to detect and understand the operating context, analyze and interpret data from more than one source, using the data to determine an appropriate action, collaborating with other agents or systems, executing actions, and improving performance by experiential learning. [2] While these agents may seem identical to human intelligence, they are not, and we can expect to have to direct them for many years to come. While there are still significant ethical issues to sort through, such as data privacy, intellectual property ownership, bias in training data and subsequent models, and transparency of decision-making processes, the potential for increased human capability and capacity cannot be ignored or wished-away.
Why Should My Company Start Using AI?
There are three reasons why your company should establish an AI practice sooner rather than later. These are:
Your team members are already using it.
If you think your employees aren’t using AI tools just because you haven’t given them access, think again. We are a species driven to finding an easier way to do a job, and the reports of people feeling overwhelmed by the size and complexity of available information are not exaggerated. It’s been less than two years since a business leader confided in me that they first suspected employees were writing their routine reports using ChatGPT when they saw an uptick in on-time submissions and a significant improvement in grammar and punctuation. Setting ground rules and encouraging team members to discuss how and why they are using available tools will help you maintain information security and build trust around your platforms.
Now is the time to direct AI progress towards community benefit.
We can’t speak cogently or offer advice about a topic or tool unless we have experienced it firsthand. In addition to building community competency, openly using and learning about AI tools within your company will create a beneficial conversation between your people about the broader ethical issues and use cases best suited to your operations. Your employees crave your leadership and encouraging them to develop knowledge, skills, and aptitude in new areas like AI will create beneficial capacity within your company, industry, and community.
Your company has more work to do than there are people to do it.
Population statistics show that U.S. labor force participation rate (the share of people working or seeking work) has been declining, and projections suggest this rate will continue to fall. [3] There are multiple reasons put forward for the decline, and in none of them does your company have an opportunity to help shift the trajectory. Therefore, you should expect there will continue to be fewer people available to do the necessary work. What you can affect is how your company uses new technologies to reduce drudgery and increase capacity. This is what some people call increased efficiency, a word which under-represents the potential for value creation once your company is recognized as a leader in developing the necessary expertise in those technologies. Like it or not, you are competing for a shrinking pool of talent, so you want to keep your current employees interested in their work and recruiting their friends and family to join them.
How Can We Get Started?
Start with tools available to you.
Many companies use enterprise-level Microsoft tools for much of their IT infrastructure. Their basic AI agents are called Copilot, and they are capable of automating repetitive tasks, answering questions, summarizing information, writing drafts, and suggesting next steps. There are other Copilot agents available that allow creation of custom AI agents based on company data and processes, and others that facilitate data analysis. Google Workspace, a popular alternative to Microsoft for smaller companies, offers similar options. Starting with existing IT platforms makes integration easier and enables experimentation.
Read the legal agreements.
Some of these tools are provided for an additional cost, which may increase sharply as you scale. It’s worth a careful review of the license agreements to understand any limitations of use and how costs may scale. Another point you should ensure is that your data will remain separate and yours, as your teams have the potential to create significant intellectual property as they develop your industry- and company-specific solutions. Don’t take the short cut here – engage legal professionals if necessary to ensure you understand exactly the pros and cons of these agreements. Share the information about what can and can’t be done with your people. They will appreciate this clarity as it will help them evaluate AI tools they might use at home.
Follow the leaders as you establish a vision for the work.
Readers may be aware of the experience of Luis von Ahn, CEO and co-Founder of Duolingo, a popular language-learning app. In April, he posted an open memo with his thoughts on AI use at the company which led to a firestorm of criticism. The original memo was removed from social media; however, Mr. von Ahn has since published a follow-up statement on LinkedIn. [5] In it, he acknowledged that he didn’t communicate his thoughts and intentions well in the first memo, and clarified his goals for how Duolingo’s people will learn and use AI tools together. It’s worth reading for his perspective and leadership modeling. People are human beings, not human doings, and we are going to make missteps.
Leadership is essential to success.
This is a scary time for all of us socially and economically. Rapid and unpredictable changes in formerly stable community pillars create uncertainty and volatility. This translates into fear-driven behaviors in the workplace. Engaging thoughtfully with your people on how the best use AI tools in your company will help you redirect their thoughts toward a constructive future. Therefore, along with developing AI knowledge and skills, you will need to develop the capacity for leadership, or the ability to cope with change. That capacity can and should exist and be valued at all levels. Set a clear direction for the company’s AI transformation. Build alignment with this vision by communicating early and often and listening as much or more than you talk. Motivate and inspire people to engage in the effort by offering support, recognition, and opportunities, and linking the vision to values meaningful to your people. Encourage networking and teamwork to find solutions to problems and to reassure your team of their essential value to and for your mutual success. [4]
The Take-Home Message
You, your team, and your company cannot afford to sit on the sidelines while others forge a path through the wilderness. Running a business is entirely about knowing what to bet on and when. While agentic AI is still in the early days, there is sufficient structure to know that significant benefits will be possible when thoughtful people come together to direct the change. Remember “The Jetsons” cartoon? We all wanted flying cars and automated conveniences, but nobody reflected on the path we would have to take to get there. Let’s pathfind to the future together!
Navigating change is challenging, and you don’t have to do it solo. I’m available to help leaders like you with strategic thinking and confidential capacity enhancement. Reach out to me for a starter conversation!
Thanks for reading Thinking Kat! You can help me by liking this post, becoming a paid subscriber, or sharing this publication with your friends.
References
[1] Encyclopædia Britannica. “Alan Turing: Computer designer.” Encyclopædia Britannica, https://www.britannica.com/biography/Alan-Turing/Computer-designer. Accessed 19 August 2025.
[2] IBM. (2025). The 2025 guide to AI agents. IBM Think. https://www.ibm.com/think/ai-agents Accessed 19 August 2025.
[3] Aaronson, S., Cajner, T., Galbis-Reig, F., Smith, C., Wascher, W., & Fallick, B. (2022, March 9). Labor Force Participation: Recent Developments and Future Prospects. Brookings Institution.
[4] Kotter, J. P. (2001). What leaders really do. Harvard Business Review, Reprint r0111f, December 2001, 3-12.
[5] von Ahn, L. (2025, June). “One of the most important things leaders can do is provide clarity. When I released my AI memo a few weeks ago, I didn’t do that well.” LinkedIn. https://www.linkedin.com/posts/luis-von-ahn-duolingo_one-of-the-most-important-things-leaders-activity-7331386411670982658-jpfX
Working with life science startup companies to solve problems, strengthen systems, and build capacity is what I do. If one of your companies or clients is struggling, reach out to me to discuss how I can help your team refocus and emerge triumphant.