#167 The AI Tool That Said No To The Pentagon.

Welcome back to On The Fly!

Here's something I don't think enough people say out loud: the tool you started with is probably not the tool you should still be using.

I’d make the bet that most people pick an AI tool, they get decent at it, and never consider making the switch to something else.

And this isn’t because it's the best one for them, but because switching feels like work.

Psychologists call it the sunk cost fallacy. You keep going because of what you've already put in, not because of what you're getting out.

I fell into that trap with ChatGPT.

However, over the last 6–12 months, I've transitioned away. I still use Chat for random questions here and there, but when it comes to deep work like content, building out an app idea, or exploring a service (something I'll share with you soon), I'm all in on another tool…Claude.

You may be wondering why I've veered away from the OG of LLMs.

That's exactly what I'm jumping into this week.

Let's go.

Alright, so why am I changing gears? I thought about it and boiled it down to 3 reasons.

1. I respect Anthropic as a company.

2. I needed a change of scenery.

3. Claude is becoming the all-everything tool.

Let's start from the top.

1. I Respect Anthropic as a Company

Recently, Anthropic (the company that owns Claude) got into a very public standoff with the US government over how its AI could be used.

Without getting too in the weeds, Anthropic had a major contract with the Department of Defense. During talks, the government pushed Anthropic to loosen some of the guardrails it had drawn around how Claude could and couldn't be used when working together. Think of guardrails as the "we will / we won't" list every company decides on when they build a product.

Anthropic didn’t want to make the adjustments it was being asked of.

The relationship broke down shortly after. You can read more about it here and here if you want the full picture.

Now, I want to be clear: I'm not getting into whether one side was right or wrong. That's a whole other conversation. What I'm talking about is business morals.

Anthropic was founded in 2021 by former OpenAI researchers who wanted safety and ethics baked into the product from day one. They literally built the company around a set of rules they wouldn't cross, no matter who was asking.

So when a multi-hundred-million-dollar contract showed up asking them to cross those exact lines…they didn't.

And that's what got me to respect the company.

Any company can slap a mission statement on their website, but they can also just as easily bend the rules around that mission statement. And it’s a whole different thing to actually stick to it when there's a massive check on the other side. Anthropic had a real shot at bending their own rules for a big payday and just…didn't. That says more about them than any product launch or flashy update ever could.

2. I Needed a Change of Scenery

Aside from the respect I have toward Anthropic, it's insanely underrated to have a change of scenery when using a new LLM.

I never thought I'd be saying this, but using Claude after years of ChatGPT is refreshing AF…especially when it comes to personal work like content, apps, and projects.

In my corporate career, I like to believe I'm extremely organized. Files in order, clean inbox, labeled folders…I take pride in all that.

But when I audited my ChatGPT, everything felt all over the place.

Tons of chats on ideas I no longer think about. Projects I made that aren't relevant anymore. Instructions that made sense at first but no longer do because I over explained. Plus, Chat referencing stuff from old chats that had nothing to do with what I was working on in a separate new chat got exhausting. I just started to get frustrated by it all.

That said, all my experience with ChatGPT actually made the switch sweeter. Because now, I’ve structured Claude in ways I never did when I first started…limiting what I put into it so it stays solely focused on my personal work. Anything random goes into ChatGPT or dare I say…Google 😱 

Starting fresh has been a game-changer. Plus, Claude has been amazing in the following areas:

• I like its reasoning

• I like how it writes

• I like their user interface

In sum: The switch has been awesome and refreshing.

3. Claude is Becoming the All-Everything Tool

What I love about Claude is they're trying to make this the home for you to operate within.

Like I've told you in past editions, we're past the point where back-and-forth in a chat is the only way to use these tools. You can have AI access folders on your computer, read them, convert them, and change the content within the file…all at your discretion, of course. (#BeSmartAboutIt)

I have an epic use case I'm sharing next week. I used a tool called "Claude Cowork" — Anthropic's agentic AI for knowledge work. In plain English: it's Claude working on your desktop, inside your actual files and folders, instead of just inside a chat window. You point it at a folder, tell it what you need, and it plans and does the work for you. So crazy.

I used it because me and my wife are moving next week and the last thing we wanted to do was write Titles and Descriptions for the stuff we're putting up on FB Marketplace. By hand, that's an hour or two. With Cowork, 15 minutes. I'll break down exactly what I did next week.

The point is, Claude's models keep improving, but the way you use them outside the chat is improving dramatically too. And that's not even getting into coding, like you can legit build websites and apps inside the Claude app using "Claude Code." Meaning, it’s never been easier to bring an idea to life online. I’ll share more about this in another edition as well.

I'm not telling you to drop ChatGPT and run to Claude. That's not the point of this edition.

What I am saying is it's worth pausing to ask yourself if the tool you're using is still the right fit for the work you're doing now. For me, the switch made sense because I learned how Anthropic acts as a company, because I needed a fresh start with what I’m working on, and because Claude is evolving into way more than just a chat window (and I’m sure ChatGPT will continue to evolve just ask quickly).

But your reasons might look completely different to me, and honestly, that's the whole point.

The truth is, none of us really know where any of this is headed, myself included, which is part of what makes this moment so interesting. What I do know is that the people who'll get the most out of the next wave of AI aren't the ones loyal to any single tool. They're the ones staying curious and willing to explore what's out there, even if it means stepping outside of what's already comfortable.

Maybe Claude's worth a look, maybe it isn't, but that's up to you to figure out. But at minimum, give yourself permission to consider something else, because a lot more is out there if you’re willing to explore.

Before You Go…

Steal this prompt idea 💡

Situation: You're considering whether switching from ChatGPT to Claude is worth it for your personal endeavors.

Prompt: Copy and paste this.

“I've been using [ChatGPT / Claude / another LLM] for about [X months/years]. My main use cases are [list your top 2–3 — e.g., writing, research, brainstorming, coding, task management].

I'm considering whether to try [Claude / ChatGPT / another LLM] instead, either as a full switch or for specific use cases.

Before giving me advice, ask me whatever questions you need to understand how I actually use my current tool, what's frustrating me, what "better" would look like for me, and what I plan on doing in the future.

Then give me:

1. An honest read on whether trying the other tool is likely to meaningfully improve my workflow, or whether I'm just chasing something new

2. A specific, low-risk way to test it this week using a real task I'm already working on

3. What to actually pay attention to during that test so I can make a real decision, not just vibe check me

Be direct. If switching wouldn't meaningfully help me, tell me straight up. I'd rather hear "stay put" than waste time chasing an upgrade I don't need.”

Try running this in whatever LLM you're using now. Then run it in one you're curious about. Compare the answers. See where it goes from there.

Alright, the plan has landed. See you next week for the Claude Cowork breakdown. It’s such a cool use case and I can’t wait to share it with you!

Find me on LinkedIn

You are now On The Fly & In The Know.