Howdy Partners, 🤠

Sorry it’s been a while since my last email, I will never send an email unless the quality is exquisite.

I've been playing with Claude's new Skills update all week, and I need to tell you about it.

This isn't another "ooh shiny feature" moment.

This will be the most important newsletter you’ll read this month.

📌 TL;DR

  • Claude Skills → Anthropic just turned Claude into a trainable employee. Package your workflows once, apply them everywhere.

  • ChatGPT Atlas Browser → OpenAI's new browser is just Chrome with ChatGPT bolted on. IMO; not worth the switch.

This update turns Claude into a trainable employee

Anthropic just shipped Skills, and it's the most practical AI update I've seen in months.

This 1-minute vid gives you an idea…

What are skills & why do they matter?

Simply: Skills are reusable instruction manuals that teach claude how you want tasks to be done, what tools to use, and what what standards/ process to follow.

You have tasks where you've figured out what works; a report structure, an analysis framework, a communication style. But every new conversation means explaining it all over again, or fixing up the AI output to match your standards.

Skills solve this.

Skills let you package your approach once, and apply it over and over in any chat.

Think of Skills as custom onboarding materials, like putting together a training guide for a new hire, except the hire is Claude.

Before Skills: You ask Claude to create a quarterly business review (QBR). It can build presentations, but matching your standards means explaining preferences each time.

You could document guidelines in a Project, but they don't automatically apply when you need them.

With Skills: Create a QBR skill once. Next time you ask for the same presentation, Claude recognises it, loads your skill (you'll see "Reading QBR skill" in its thinking), and generates a deck matching your standards from the start.

Where Skills fit with how you already work:

Projects = accumulated context for ongoing work.

Skills = repeatable procedures applied automatically everywhere (regular chats, inside Projects, across all your work).

Use both together when you need accumulated context and consistent standards.

Skills you should be building:

  • Brand guidelines & voice → so every output matches your standards without checking

  • Ad creative brief generator → Your framework for briefing designers/agencies with target audience, key messages, CTAs, and brand guidelines built in

  • Content repurposing workflow → Takes one piece of long-form content and spins it into tweets, LinkedIn posts, email, and short-form following your distribution playbook

  • Viral social media post writing → your tone, format, and hooks packaged for any platform

  • SOP documentation template → Your exact format for writing standard operating procedures that actually get followed by your team

Create a skill when you've figured out how you want something done and want Claude to follow that approach automatically.

Business opportunity worth noting:

This opens a market for packaged skills as digital products.

Think: viral X post writing skill, pitch deck creation skill, SEO blog post skill.

Skills come in the form of a .skill file, so skills can be sold, used as lead magnets, or given to clients.

If you've refined a repeatable process that produces results, you can now package it as a skill others can use.

How to start building your own skills:

Watch my Claude Skills 101 video below!! I’ll show you how to create your own custom skills from scratch, step-by-step (takes 10 minutes).

Plus, grab my free viral hook writing skill - trained on 100+ proven ad templates and psychology-backed copy patterns from the world’s best advertisers.

Use it to make your AI write scroll-stopping, curiosity-sparking hooks that stand out anywhere you post.

OpenAI just launched Atlas - their answer to the AI browser

The Atlas browser is ‘Chromium Based’, meaning it is literally just google chrome, but with ChatGPT in it. If you've used Dia or Comet, this will feel familiar.

You get an "Ask ChatGPT" sidebar that pulls context from whatever tab you're on, plus an Agent mode that can click around and complete tasks for you.

This feels like ChatGPT wrapped in a browser skin, not a meaningful upgrade.

Agent mode works exactly like it does in the regular ChatGPT app, only handles one tab at a time, runs painfully slow, and breaks on anything beyond basic web pages.

When you search, it defaults to ChatGPT instead of the web - which just means you're using ChatGPT's web search feature with extra steps.

For now, stick with your regular browser and pull up AI when you need it. Atlas doesn't add enough value to justify the switch.

💡 Prompt of the Week (BONUS)

Make ChatGPT or other LLM think more independently rather than just agreeing with you all the time. Use it when you’re doing thinking work where being wrong is costly or uncertainty is high, especially early in the process.

Going forward, avoid simply agreeing with my points or taking my conclusions at face value. I want a real intellectual challenge, not just affirmation. Whenever I propose an idea, do this:

Question my assumptions. What am I treating as true that might be questionable?

Offer a skeptic’s viewpoint. What objections would a critical, well-informed voice raise?

Check my reasoning. Are there flaws or leaps in logic I’ve overlooked?

Suggest alternative angles. How else might the idea be viewed, interpreted, or challenged?

Focus on accuracy over agreement. If my argument is weak or wrong, correct me plainly and show me how.

Stay constructive but rigorous. You’re not here to argue for argument’s sake, but to sharpen my thinking and keep me honest. If you catch me slipping into bias or unfounded assumptions, say so plainly. Let’s refine both our conclusions and the way we reach them

🛠 Tools to Try

  • Brain.fm → Focus music that actually works. I've never been so locked in. Try it if you need to block out the world and get shit done. Fun fact: I made a Spotify playlist called "Brain.fm" a few months back and they sent me a cease and desist letter 😂.

  • Anthropic's Official Skills Repository → Plenty of pre-built skills here you can steal and start using.

  • Notion Skills for Claude → A repository created by Notion full of Notion skills.

  • Free viral hook writing skill → grab my free viral hook writing skill

🍲 Brain food

  • Dan Koe’s 30 day dopamine detox → Started mine 10 days ago. No social media, no coffee, no junk food. Feeling fucking amazing. Also explains why this section's a bit light this week. I haven't been consuming much of anything.

  • curated.supply → When I make it, I won't tell anyone - but there will be signs

  • Headache on SpotifyI've been bumping this artist recently while working. their music is so strange, but there's something kind of nice about it.

  • Ultimate AI video model comparison → Sora 2 vs Veo 3 vs Kling 2.5 vs Wan 2.5

Alright, that's a wrap for this week.

My car got broken into this morning and they made off with my MMA gear, so if you see someone trying to sell slightly worn gloves on Facebook Marketplace... that's probably mine.

Nonetheless, it's going to be a great week. I've got a dozen Claude Skills to build and zero patience for re-explaining my workflows ever again.

Catch you next week,

Remy :)