Today Claude tried to be helpful.
It failed spectacularly.
I uploaded specific files about NEO and Augmented Intelligence.
Asked it to reference those files for a content strategy.
Should have been straightforward.
Instead of looking at the files I gave it, Claude went rogue. Searched my entire Google Drive.
Pulled documents from old employers. Past businesses. Dead projects. Years of evolving thoughts all mixed together like some digital archaeology experiment.
Then it confidently presented this mess as if it’s the current state of my business.
I spent the next ten minutes untangling context pollution from 2019.
The worst part? There's a toggle to turn off Drive search. But it's on by default.
Because Anthropic thinks searching everything is "helpful."
In December 2024 I was singing praises to Anthropic.
These days, I’m feeling increasingly meh about them.
Not only them, but I’m concerned about where we’re heading, industry-wide.
Every new AI feature just seems intent on removing moments where you'd normally use your brain.
Claude searches everything so you don't have to think about what to search.
It gives nine answers so you don't have to think about which one you need.
Ask why it did something, it immediately rewrites everything so you don't have to think about what to change.
They put in Gmail integration so you don't have to remember what you wrote.
Calendar search so you don't have to track your commitments.
Cross-conversation memory so you don't have to recall previous discussions.
Each feature sounds helpful in isolation.
But together, they're only creating learned helplessness at scale.
And they're trying really, really hard to do this.
(Even if they're getting most of it wrong…for now)
The developers are so focused on removing friction that they're removing the need to think.
So many people are debating AI safety. Worried about robot uprisings and AGI overlords.
But the real threat is cognitive extinction through hyperconvenience.
The apocalypse won't be robots taking over. It'll be humans forgetting how to think.
We'll be so dependent that if the servers go down, we'll simply be paralyzed.
Unable to write any emails.
Make any decisions.
Solve any problems.
That’s where the entire industry, basically, is leading us to.
The irony is they can't even do this properly.
Example: Six months ago every AI bro was screaming about AI agents.
"2025 will be the year of AI agents!"
"Your business will run itself!"
"Fully automated everything!"
Now it's almost the end of 2025. Where are these magical agents?
Nowhere. Because AI can't even search the right folder without adult supervision.
They promised us autonomous businesses.
We got Claude searching my 2019 tax documents when I asked about NEO.
They're incompetent at making us incompetent.
But they keep trying. Adding more features. More automation. More "help."
All while we're hitting usage limits every other hour and getting "server overwhelmed" error messages.
The industry is so busy trying to build AI that thinks FOR us that they're breaking the tools that help us think WITH them.
This is why I'm building NEO differently.
No automatic searches. No unwanted suggestions. No trying to be smarter than you.
But clear thinking, enhanced by AI, directed BY you.
Because augmented intelligence means making YOU more capable, not more dependent.
Making AI your precise tool, not your digital overlord.
The fitness industry tried to sell shortcuts. People got weaker.
The coaching industry tried to remove struggle. People stopped growing.
Now AI is trying to remove thinking.
You already know where this leads.
FELIX
P.S. They just announced they'll train Claude on our conversations by default starting September 28.
Unless we opt out.
As if context pollution wasn't bad enough, now they want to train Claude on millions of mediocre conversations.