July 1, 2025

”I can’t tell management that this AI coder doesn’t actually work because I’m afraid the CEO will cut my job.” — Overheard by a software engineer on the streets of San Francisco

To hear VCs tell it, AI prowess knows no bounds. Need engineers to write code faster? AI can do that. Need more outbound calls? AI can do that. Want a better pitch deck? AI can do that. Need to interview more candidates? AI can do that. Improve our medical diagnosis or understanding? AI can do that.

Someday this VC-fueled future may arrive, but as a CEO today you should be more skeptical. VCs need to think 5-7 years ahead, and make a killing when they bet right one out of ten times. As a CEO you need to focus on a nearer time frame and produce more reliable outcomes.

Fool Me Once…

Like most technophiles, I leverage AI daily. I vibe code with Cline, summarize interview transcripts with Claude, and leverage ChatGPT to interpret medical lab reports. AI has become a ubiquitous tool in my arsenal and I have no doubt that one day it will [REDACTED].

Also like most people (at least when they’re not spouting on social media), I’m frequently bumping against AI’s current limitations. In code, I caught Cline trying to add a secret key to my repo. ChatGPT way overestimated the severity of my medical report, citing hallucinated numbers. In summarizing interviews, Claude quoted insights that were never mentioned. The list goes on. Whenever I point out its errors, AI flatters me with a “great point!” and simply changes its answer.

Despite LLMs so often missing details or getting important things wrong when we take the time to examine their answers, we still turn to a new chat and trust it implicitly on areas where we don’t bother to verify. This is a modern form of Gell-Mann Amnesia, in which we read one newspaper article about an area of personal knowledge and chuckle at all the things it gets wrong and then turn the page and take on faith what it reports.

Why would you trust AI to run a key component of your business when so often it delivers amazing demos but only mediocre results?

Introducing The AI “Pragmatic Optimist”

Current challenges notwithstanding, it’s clear that AI has already become a powerful tool, and will only get better with time. A CEO in 2025 can no more avoid AI than one in the early 2000s could avoid SaaS or the internet. Like these other platform shifts, AI will (eventually) transform how we create, deliver, and consume software. Every CEO must at some level be an AI Optimist or risk their company lagging behind. (Or even worse, offending our future overlords; how’s that for optimism?!)

So how does a CEO take advantage of AI’s current possibilities, position their company for the future, while not destroying the trust from important stakeholders — buyers, users, employees — that will be essential to its long-term success?

The key is to view AI today as primarily a productivity tool and not a magic wand. When looking at near-term strategies and viewing it as a tool, the pragmatic CEO needs to look at ROI, outcomes, and how to drive adoption. This playbook resembles a classic innovation change management recipe: paint a vision, align on business goals, move fast in small increments, and build momentum with compounding wins. Given the fast moving and fragmented nature of the market, AI platforms and tools are likely to change and warrant a healthy skepticism.

The rational CEO will act as an optimist and begin to adopt AI while adopting pragmatic outcomes rooted in attainable productivity gains. This approach allows for near-term benefits, positions the company to accelerate as LLMs improve, reduces the risk of major re-work as the market shifts, and de-risks the business against the upcoming platform shift.

Conclusion

There’s no need to chase every shiny demo or wait for the perfect use case. The real opportunity lies in applying your judgment—grounded, skeptical, and optimistic—to identify where AI can actually move the needle for your business. That means testing ideas, measuring real outcomes, and scaling what works. If you’re looking for a partner to help you navigate this shift with both eyes open, we’d love to talk. TCGen can help you explore what’s real, what’s hype, and where to focus next.

Key Takeaways

  1. AI is powerful but still unreliable — beware of Gell-Mann Amnesia
  2. AI today is best seen as productivity tool, and adoption should follow classic change management
  3. But it’s also a platform shift, and it’s important to invest now