Deutscher

If you could have a call with the #1 AI expert in the world, what system would you have them help you set up?

Claude runs my entire life, and it's the best investment I've ever made.
Ideation/creative tasks = Sonnet 4.6 + Opus 4.6
Vibe coding = Opus 4.6 on web/app
Heavy coding tasks = Opus 4.6 + Claude Code CLI
Scheduled tasks = Cowork
Repetable workflows = Cowork Plug-ins + Claude Skills
Gym advice = Sonnet 4.6
Nutrition coaching = Sonnet 4.6
Quick queries = Haiku 4.5
Claude Chrome extension = Haiku 4.5
Claude x Slack (and other tools) = Sonnet 4.6 & MCPs
What else gives you THIS much ROI for $100-200/month?

"i dOnT kNoW wHaT tO aUtOmAtE"
That's not an excuse anymore.
Here are 30 *realistic* automation ideas to 10x your daily productivity.
With the exact tool options included.
Pick a few and stop overcomplicating it.

My march 2026 AI stack as an AI-addict (in order of usage):
• Claude (writing, coding)
• GPT (brainstorming, strategy)
• Grok (live updates, markets)
• Manus (agentic workflows)
• Gemini (photos)
• Perplexity Computer (multi-LLM workflows)
• NotebookLM (learning, transcribing)

The best use case for AI is brainstorming - bar none.
• Content
• Business problems
• Roadblocks
• Blindspots
Even if you don't do any agentic shit, and you just use it to brainstorm - you'll benefit massively.

Are anyone else's LLMs (Claude & GPT) absolutely fucking stupid today?
The quality of responses are much much worse than usual.
Must be something to do with conserving energy/data centres.

This Dubai stuff is fucking nuts.
This has to stop.
Stay safe everyone.

Stay safe guys. 🙏

This is f*cking wild.
The Pentagon just blacklisted Anthropic, labelling them a "national security risk".
Trump called them “leftwing nut jobs", making a “DISASTROUS MISTAKE.”
Just hours later, Sam Altman announced OpenAI had struck a deal to replace them.
The crazy part is OpenAI’s deal has the exact same terms Anthropic was just punished for...
Here’s the full timeline:
• July 2025: Anthropic signs ~$200M Pentagon contract. First AI cleared onto classified networks.
• Jan 2026: Defense Secretary demands models be usable for all lawful military purposes.
• Anthropic agrees broadly - but sets

The irony is
People are trying to learn AI to avoid unemployment
But AI is moving so fast, the only way to keep up is to be unemployed

A lot of AI is a jerk off.
You spend hours setting it up/experimenting - and by the time a workflow is working, a better alternative ships.
The solution is to be super intentional about WHAT you're automating and WHY you're automating it.
If you can't prove that it will save you hours of time, don't make the time/financial commitment to automating it.
(side note: there IS definitely some value in experimenting purely to up-skill/test new products - and I encourage it - but this needs to be budgeted as creative time and not prioritised over other things where the ROI is more clear).

If you're still using ChatGPT, but don't have a Claude subscription - I don't know what to say to you.
You're missing out big time.