Welcome back to the AI Edge weekly newsletter!
This week, we have a special one for all Claude and Codex users!
If you’re using these tools, we’re convinced that this week’s edition will completely change how you use them.
Anthropic and OpenAI have recently shipped “/goal,” a new AI slash command that lets you run AI agents for 8+ hours with zero manual intervention.
You set a goal, and your agents run autonomously until that goal is complete.
This new feature is a complete game-changer, and in today’s publication, we’re going to show you exactly how to use it.
By the time you’re done reading, you’ll be able to launch AI agents that work while you sleep (literally).
After our main anchor piece, be sure to stick around for the Midweek Edge, where we recap the top AI news from the past week, and our Looking Ahead section, where we break down top AI trends & research.
Enjoy!
What exactly is /goal?
As mentioned above, Anthropic and OpenAI both recently shipped /goal - a slash command that lets an AI agent work in a loop to complete tasks without needing your permission at every step.
If you use Claude/Codex regularly, you know the feeling:
You prompt, the agent acts, you approve, you prompt again, and that cycle repeats every few minutes, defeating the whole point of having an agent.
With /goal, that loop closes.
You set the goal once, and the agent validates its own progress, makes its own decisions, and keeps working until the task is done. No check-ins. No approvals. No babysitting.
Think of it as telling an employee to complete a project and come back only when it's finished.
I understand what /goal is - now, how do I actually use it?
How to use /goal (deploy agents)
Using /goals is extremely simple.
Getting started in Codex:
Open Codex desktop and go to Settings → Configuration and ensure goals = true
Start prompting /goal prompts
Example: “/goal I’m going to sleep - conduct some research for me on all the top AI newsletters and compile that information for when I wake up.”
You’ll know you did it correctly when a check mark appears to confirm the goal is set ^
Getting started in Claude Code:
Launch Claude Code CLI and start prompting with /goal directly.
That’s it! You literally just have to launch Claude Code in your CLI and start typing /goal prompts.
It really is that easy.
Now that you know how to use /goal, let’s cover effective /goal prompts:
Writing Effective /Goal Prompts
At its core, your /goal prompts can be as simple or as complex as you’d like.
Here’s some general advice so you don’t wake up to slop and actually get AI results you want.
We recommend including at least three things in your /goal prompts:
Your goal: your task/goal in one line
Measurable end state: define what done actually looks like
Constraints: rules and constraints the model must abide by
Putting these three things together, your /goal prompts should look like this:
Prompt template
/goal [do the work] until [measurable end state] without [constraints that must hold]
Real example
/goal fix every failing test until npm test exits 0 without modifying any file outside the /auth directory.For more complex tasks, we recommend using this advanced prompt structure (save this):
Pro Tips
Pair /goal with /plan. This is how you get the best possible end results.
Use it for non-technical work. Research, writing, docs, file organization - anything with a clear end state works.
Give your agent a .md tracking file. It'll reference it throughout the run and track all its actions for you to view once completed.
Let it run overnight. Set your hardest, most complex task before bed. Come back to a finished product in the morning.
Let the agent write its own /goal prompt. Describe what you want in plain English and ask it to generate the prompt. It’ll probably write a better one than you.
That’s /goal in a nutshell.
Set your /goal and let your Claude/Codex agents run while you sleep (also works in Hermes agent too).
Now, let’s jump into the Midweek Edge - our manually curated list of only the most important AI updates to ensure you stay up to speed.
🚨 Google I/O Rollout Begins🚨
New Gemini app
Gemini 3.5 Flash LIVE now
Gemini Spark (beta) rollout begins this week
Gemini Omni (media model) LIVE now
Explore the Google AI ecosystem to try the new Google I/O tools!
🚨Pope Partners With Anthropic?🚨
Ok, this one is pretty crazy.
Pope Leo XIV released a massive, 42,000-word encyclical (a formal papal document) warning the world about the dangers of artificial intelligence.
While sitting down with the Anthropic team, he announced they will work together to “find the way for humanity.”
🚨Uber Burns Through Its Entire 2026 AI Budget In 4 Months🚨
You read that right.
Uber has already burned through its entire 2026 AI budget.
The COO stated AI productivity may not have been what they thought, resulting in spending cuts.
Amazon and Microsoft have also announced similar concerns about AI expenses.
🚨OpenAI Files For IPO? ($1T+ valuation)🚨
OpenAI is rumored to confidentially file a draft of its IPO as soon as this Friday.
The target listing window is September to November 2026. OpenAI has a current private valuation of $852B from the March round.
Analysts expect the public valuation to push past $1T.
Looking Ahead - The top AI trends, leaks, and news we’re closely monitoring.
👀 Gemini 3.5 Pro - Expected June 👀
The world’s most powerful model is set to release in June - Gemini 3.5 Pro.
Hopefully, it’s better than what 3.5 Flash has been!
👀 Grok Build Beta 👀
Grok Build Beta is live now - as mentioned last week.
It is now available to all SuperGrok and X Premium+ users.
BONUS: How to Automate Anything in Your Life (3 steps)
Final Thoughts
If you made it this far, thank you for reading, and we hope you found this week’s edition valuable.
If you did, please share this with someone who you think would benefit from our publication.💙 Our promise to you: Every Wednesday, at 7 am EST, we’ll cut through the AI noise and send you human-curated AI content, tool guides, workflows, and more to make sure you stay ahead of the AI curve - straight to your inbox and 100% free. |
Recent & upcoming content:
YouTube Content
Live Now: “I don’t trust Claude anymore…”
Upcoming Content: “I Fired 5 Employees & Replaced Them With AI!” - LIVE SOON
X (Twitter) Content
Live Now: “How to Automate Your Life”
Upcoming Content: “These Claude Prompts Changed My Life” - LIVE SOON
Instagram Community & Free AI assets
Live Now (early access for all newsletter readers): I Replaced 5 Employees With AI (full cheatsheet to all five AI workflows)👇
For all the AI prompts, cheatsheets, and PDF guides mentioned on the AI Edge YouTube channel, grab them by joining Miles’ free Instagram community here: |