Reading time: 4 minutes
Greetings from above,
Two years ago, asking AI for help was like hiring a consultant who only gives advice and never picks up the phone. This week, that consultant started actually making the calls.
I spent last Tuesday doing research the old way — fifteen browser tabs, a half-built spreadsheet, and that slow creeping feeling that I was doing a robot's homework for it. Then I tried the new approach. It just went and did the work. No hand-holding. Took me a minute to trust it.
Today we're covering:
What actually changed with Claude Sonnet 5
The pricing window that's really a countdown clock
One prompt you can paste right now to see it work
Let's get into it.
What Actually Changed
For two years, AI could tell you how to do something. Now it does it. Anthropic released Claude Sonnet 5 this morning, and the difference isn't a smarter chat window — it's an operator.
Give it a goal, and it builds a plan, opens a browser and a terminal, works through the steps, and hands you a finished result. You don't manage it. You review it.
How This Changes Your Workweek
It researches, plans, and executes instead of just describing what to do
It's priced low right now, so using it often actually makes sense
"Write me an email" becomes "build me a campaign"
From Tasks To Missions
Here's the real shift. Yesterday, you'd ask for an outreach email, then still do the research, build the list, and hit send yourself. Today, you can hand over a goal like
"find 20 agencies that just raised funding, write each one a personalized opening line, draft the emails" — and it goes and does it.
Here's a prompt that shows you the difference. Paste it, give it a real goal, and watch:
You are my autonomous operator, not a chatbot.
Mission: [PASTE A REAL GOAL — e.g. "find the 5 fastest-growing competitors in my niche, pull their pricing and positioning, build a one-page teardown showing where each is weak."]
Rules:
- Work in steps. Before each one, say what you're doing and why.
- Use the browser to verify — never guess.
- Make the calls yourself and explain them.
- Deliver the finished asset, then give me a 3-point list of what to double-check.
Run it once. You'll stop typing "please write about..." after that.
The Step-By-Step
Open Claude Sonnet 5 and paste the mission prompt above
Swap in a real goal — one with an actual deliverable, not just an answer
Let it lay out the plan before it starts
Watch it verify facts through the browser instead of guessing
Review the finished asset it hands back
Check the double-check list it gives you at the end
What This Means For You
Stop describing tasks. Start assigning outcomes and let it own the steps.
The gap between people who use AI and people who run it is about to get wide, fast.
Lower pricing right now means more room to experiment without watching the meter.
Wrap Up
What you learned today:
Claude Sonnet 5 plans and executes on its own — it's not just a better chat window
The leverage now lives in the mission you assign, not the prompt you write
Low cost plus autonomy is the combination that actually frees up your time
The tools that used to just talk are starting to work. That's a real edge, while it lasts. Go give one a real mission today.
Then hit reply and tell me what it built. I read every one.
Thanks for being part of this community,
Keep learning,
🔑 Robert from God of Prompt
