The next software bubble might not be valuations.
It might be supply.
Millions of apps.
Millions of tools.
Millions of agents.
And a finite number of humans willing to care.
There’s an uncomfortable future where employers expect everyone to use AI…
but still judge you when AI makes a mistake.
All the productivity upside.
All the responsibility stays human.
Consumers say they hate subscriptions.
Then subscribe to:
- Netflix
- Spotify
- iCloud
- YouTube
- five other random apps
The problem isn’t subscriptions.
It’s subscriptions that don’t earn their place.
Job posts still want five years of React, system design theater, and culture fit.
Then the first week on the job is “here’s Cursor, don’t paste secrets, ship.”
The obsession with originality is ridiculous. Customers do not care whether you invented the category. They care whether your version is better for them.
2,200 followers.
1.7M impressions.
47K engagements.
26K likes.
10K replies.
And almost all I've done is talk about:
AI replacing developers
building stupid apps
building useful apps
startup delusion
software as a whole
and the increasingly weird future we're heading into.
Turns out having something to say beats having a content strategy.
Agree or disagree?
Announcing Gemini 4 Argon, our new frontier model.
Argon is built to sustain deep reasoning across complex, long-horizon workflows and delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.
We’re also expanding the model’s output token limit to an industry-leading 1M tokens.
Argon is currently rolling out to a set of trusted cyber defenders in the Fairwind Program, with broader availability as soon as possible.
So after all the hype around OpenAI DevDay...
We got:
- a cheaper model that still doesn't beat Opus
- usage limits cut in half
- and a new $500/month tier if you want substantially more compute (that still runs out super fast)
Meanwhile Anthropic is still leading in benchmarks and actual real world tests.
At what point do we admit OpenAI dropped the ball?
Because cheaper inference + lower allowances isn't the AI progress I think most people were expecting.
Opus 5.5 is still the GOAT.