Recently, I came across Anthropic’s fascinating experiment, Project VEND-1, where they gave Claude AI $1,000 to start and run its own business. Spoiler: the AI lost money.
Some people laughed it off, but it reminded me of something I’ve cautioned clients about for years: the misleading comfort of so-called “best practices.”
When I worked in development transformation consulting, I often warned teams about the dangers of blindly following “best practices.” The name sounds like a blueprint for excellence, but often, it simply means “common practice.”
I used to tell teams bluntly:
“Doing what everyone else does will give you the same results as everyone else.”
And that’s not necessarily a bad thing. Sometimes you want stability and predictability. But it’s crucial to be intentional. You shouldn’t assume “best practices” will produce superior results, especially in areas where above-average outcomes are critical to your business success. That’s where creative thinking, tailored experimentation, and unique approaches become essential.
Which brings us back to Claude.
Large language models like Claude, GPT-4, and others are trained on vast datasets pulled from across the internet. They don’t know which ideas are brilliant and which are just recycled conventional wisdom. They’re designed to predict what the most likely, average answer might be in a given context.
This is incredibly useful if you’re:
In these cases, the “average” is a huge step up. AI can save time, provide summaries, and explain concepts clearly and quickly.
But there’s a flip side.
Some decisions are too important to settle for average.
When you’re building a business, launching a product, designing strategy, or building something disruptive, average results are often simply not good enough.
And here’s the rub: the average “business advice” floating around online is, by definition, a blend of successes and failures. It’s important to remember that:
(Source: U.S. Bureau of Labor Statistics)
That means a substantial portion of the data used to build LLMs, comes from companies that didn’t make it. An automated business venture losing money isn’t shocking. It’s almost inevitable. Failure is an incredibly human and incredibly common outcome in business.
Anthropic’s experiment with Claude is a timely reminder: AI is great at synthesizing the average, but the average is often mediocrity.
If you’re leading a business, especially one aiming to stand out, you need to go beyond default outputs. Use AI to support efficiency, speed, and ideation. But never let it replace your judgment, creativity, or strategy.
Because if your AI strategy looks like everyone else’s you’ll likely end up where most companies do:
The middle of the pack. Or out of the game.
The takeaway? AI can be a force multiplier, but only if you pair it with strategic intent. Following the average will only get you average results. If you want to outperform, you need to think beyond what AI, or “best practices” suggest by default.
The businesses that win will be the ones that combine AI’s scale and speed with human insight, creativity, and bold decision-making.
If you’re serious about building something that outperforms, not just conform—now is the time to act.
Join us for a 90-minute AI ROI Discovery session, where we’ll identify where AI can create measurable, differentiated value for your business.
No fluff. Just real use cases, tailored to how your teams actually work.
Author – Paul Spears (Head of AI, Valiantys) https://www.linkedin.com/in/paul-spears-97833521/
Register for a session and start turning AI from theory into traction. Contact us for more details.