Deutscher

CANCEL your plans tonight.
If you're not automating your life with AI, bookmark this.
A list of EASY ways to start automating your life with AI:
• Claude Skills (if you work on a computer, I guarantee you can build at least 5 useful agentic skills)
• Scheduled tasks in Claude - for any repetitive daily tasks.
• Second brain system. Point Claude at an Obsidian vault and let it self-update as you add meeting notes, general notes, etc.
• A local memory folder. A few Markdown files with your goals, preferences, and context. Good for swapping between AI tools.
• Loops. /loop inside Claude Code for

If you're trading crypto/stocks, the highest-leverage action you can take right now is to start learning how to incorporate AI into your trading processes.
Think: backtesting strategies, spotting low-caps, data analysis, live market data reports, etc.
Even simple workflows like building custom trading journals are complete game-changers.
Over the past year, AI has gotten so good, and if you haven't tested the latest models (Fable/Sol), you're seriously missing out.
Admittedly, there is a bit of a learning curve, but once you get over the initial hump, you will never go back.

One of the biggest problems with AI right now is the lack of synced memory.
You use Claude for one thing, ChatGPT for another, and none of the context is synced.
I came up with a simple system that solves this, and if you use multiple AI tools, you should set this up.
The core of the system is a local desktop storage vault.
The setup (a few minutes max)
Create a local folder on your desktop with Markdown files that hold everything you want any AI to know about you.
Some things to include: your business context, your goals, your preferences, your ongoing projects (you can export things from you

So with Grok 4.6 you're basically getting:
• Fable 5 at 80% discount
• GPT-5.6 Sol at a 60% discount ($2/$6 vs $5/$30 per million tokens)
• Faster response than both
The only caveat is a 500k context window & Fable still seems to be the pound-for-pound coding king.
Regardless, still an insanely impressive launch.

If you have no clue where to start with AI, I HIGHLY recommend building Claude Skills.
Hands down, the lowest-hanging fruit for getting started with AI automation.
This is my favorite Claude Skills prompt (guides Claude through building helpful agentic skills for you):

One of the highest-leverage productivity unlocks anyone can do, not only for their business but also for their personal life, is to set up a second-brain system.
If you're anything like me, you've probably tried Notion, Obsidian, or some version of an AI second-brain setup.
While these tools are definitely functional, I built a system that is 10x better.
The real problem with these tools is what happens the moment you stop feeding them.
Stop adding notes to Obsidian, and it goes stale. Stop updating your databases in Notion, and it stops being useful.
Every one of these systems runs entirely o

If you don't know how to do something, literally just ask Claude Code.
We live in the greatest era ever for curiosity and execution.

I've used Claude to backtest, build, and deploy dozens of AI trading bots.
This prompt is absolutely cracked.
Paste this into Claude Code, and your agent will guide you through the exact process for building your first trading bot.
Traders need this:

If you're struggling to make progress in your life, read this post.
(financially, business - whatever)
This one habit has changed my life more than anything else, and I haven't seen anyone else talk about it.
This single mindset shift has helped me exponentially compound:
You need to treat your brain like an algorithm.
The more data you feed it, the faster it iterates.
Instead of reflecting every year, why not reflect every quarter? Every month? Every week? Every day?
I reflect every single hour (maybe a bit overkill).
Not in a stressful way, but just quick check-ins to make sure every single

I think one of the best ways to become AI-native is to build "mini tools" for yourself.
Examples: a meeting note-taker, a custom notes/journaling app, a prompt optimiser skill/dashboard, and so on.
None of these require you to know how to code, but they do require you to learn Claude, iterate, learn context engineering, basics of vibe coding, etc.
The real value isn't the tool itself because, frankly, you can just pay a subscription for some SaaS.
But the actual process of building yourself is hands-down the fastest way to become AI-native.
Beats watching YouTube/consuming content any day.

The entire timeline is complaining that Sonnet and Opus 5 are dumb.
I don't blame them. Sonnet 5 and Opus 5 are dumb out of the box.
Fix a few things, and they instantly feel 10x smarter.
Thank me later (advice straight from Anthropic:
1. Stop porting over old prompts
Skills and system prompts built for older Claude models can actually make Claude 5 worse, not better.
Anthropic says to start fresh instead of reusing what worked for Sonnet/Opus 4.6 or older.
2. Stop telling Opus 5 to "double-check itself"
It already catches and fixes its own mistakes without being asked. Adding "verify your ans

No one is actually making money with vibe-coded apps.
Change my mind.👇

AI is the greatest money-making opportunity you'll get in your entire lifetime.
Lock in now, and you could completely change your life by 2030.
There are 4 steps to printing money with AI - and if you want to make your first $10k with AI, this is the roadmap to follow:
Step 1: Building Memory & Context
Nothing else matters until AI actually knows about you and your goals.
You NEED to build a local memory folder that holds your business strategy, brand identity, and monetisation plans.
Feed it constantly. Meeting notes, brain dumps, documents, so every new AI session already knows everything.
S

General consensus is that Opus 5 is the worst Anthropic model release ever.
We went from Fable 5 euphoria to pure despair in a matter of weeks.
No other space like AI.

The #1 question in my DMs every single day:
"Miles, I see your content, but how do I actually start making money with AI?"
My answer is always the same, and it surprises people every time.
Getting started with making money with AI has nothing to do with tools, niches, or even business ideas.
It's all about memory. Let me explain:
AI is simply a layer you slap on top of your existing data to become more efficient, or in this case, monetise.
If AI doesn't actually know anything about you or your skills, nothing else you try to build on top of it matters.
Imo, building a solid memory system isn't

If you trade financial markets and are not using AI regularly, you're doing yourself a massive disservice.
I don't say this lightly. I've been investing for 8 years, and AI is the single biggest edge shift I've had in my entire investing career.
Most traders are still doing research the way people did it 10 years ago.
Manually scanning news, manually pulling financial statements, manually backtesting strategies, and so on.
Meanwhile, the people actually using AI properly are backtesting strategies in minutes, getting instant earnings breakdowns filtered through their own thesis, and running re

AI tip to cut your token costs by literally 80%:
(works in Claude, Codex - literally any harness/model)
Barbell strategy + subagents.
Here's the idea:
Most people run every part of a task through the same model.
Let's say they use Fable 5 for planning, execution, and review. This is probably the most wasteful way to use these frontier models right now.
I've been using a barbell approach that splits the work by intelligence needs.
Heavy on both ends, light in the middle
→ Planning (10%): use your smartest, most expensive model to architect the task, define the spec, and set the success criteria