Apple’s AI strategy
Apple’s new CEO says they won’t ship AI for AI’s sake.
The new Siri runs on Gemini.
The China version runs on Alibaba’s Qwen.
Apple’s AI strategy is to ship everyone else’s AI and not make a big deal about it.
It might be the smartest play in the industry.
No os miento cuando digo que quizá nos encontramos ante la mayor cuenta de Twitter de la historia relacionada con la IA
No he visto un tweet malo de esta cuenta
DAY 6 OF 30: ONE PERSON PLUS AI CAN DO THE WORK OF A SMALL TEAM
A small content team used to need 5 roles:
1. Researcher
2. Writer
3. Editor
4. Designer
5. Analyst
Now one operator can cover all five, with AI as the support for each role.
This is not about replacing the team. It is about one person doing the full loop, fast.
HERE IS HOW EACH ROLE CHANGES:
Researcher: Claude gathers and summarizes sources. You check the facts.
Writer: Claude drafts from your brief. You add the voice and the point of view.
Editor: Claude flags weak sections. You make the final call.
Designer: AI tools create the first visual. You pick the best one.
Analyst: Claude reads your numbers. You decide what to test next.
The rule: AI does the first 80%. You own the last 20%.
That last 20% is judgment, taste and accuracy. It is the part that cannot be outsourced.
DO THIS TODAY (45 minutes):
1. Pick one piece of work your team, or you, produces every week.
2. List every step from idea to publish.
3. Next to each step, write who does it today, and how long it takes.
4. Paste the list into Claude with this prompt:
"Here are the steps in my workflow and how long each takes. Tell me which steps AI can do for me, which need my judgment, and which should stay fully human. Give me a new workflow."
5. Run the new workflow once on a real task.
6. Write down the time before and after. Use real numbers only.
7. Save a screenshot of the old workflow and the new one.
Your before and after is the proof. Do not exaggerate it.
AI does not remove the team.
It removes the waiting between the steps.
Tomorrow: how I went from crypto and content to AI-native operations.
Anthropic’s “training a competitor” flag
A PhD student says Anthropic flagged his account for “training a competitor.”
His reinforcement learning jobs were Claude-assisted.
So where is the line?
If you use Claude to do research, you’re fine.
If you use Claude to build something better than Claude, you’re a competitor.
Every lab wants you to build with its model and nobody wants you to beat it.