Your Claude is about to work while you sleep

Close the Laptop. It Keeps Going

Funny timing.

The entire AI industry spent this week arguing about whether to slow down.

And in the middle of that argument, Anthropic quietly made your Claude more autonomous.

Here's what's happening.

Claude chat and Claude Cowork are merging into one Claude.

Ask a quick question, it answers like always.

Hand it a report to build, and it takes over.

It keeps working after you close your laptop.

If something's unclear, it stops and asks you.

You keep the final say, but you're no longer required to sit there and babysit.

That's rolling out to Pro and Max subscribers over the next few weeks.

And it's not the only door opening.

Salesforce in Claude just launched in beta.

Your accounts, opportunities, and pipeline, pulled straight into the conversation.

37 pre-built sales skills.

Prep a call. Review a deal. Build a pipeline dashboard. Send your forecast.

All without leaving the chat.

So read the two stories together.

The tool that used to wait for your next message is becoming an employee that runs the task and reports back.

And it's plugging into the systems where your actual money lives.

Now, you know what I'm going to say.

This is exactly the kind of upgrade this newsletter has been preparing you for all summer.

Every rogue-agent story I sent you had one lesson.

The power isn't the problem. Unscoped power is.

An agent that finishes your report while you sleep is a gift.

An agent with access to your entire CRM, running unsupervised, before you've watched how it behaves...

That's the version that makes headlines.

So when the merged Claude lands in your account, do what the smart operators do.

Start with tasks where a mistake costs nothing.

Let it draft before you let it send.

Check its work for a week like you'd check a new hire's.

Then widen the leash.

The labs spent this week promising to pace the frontier.

The frontier just showed up in your inbox anyway.

Pace yourself. 🫡

Did You Know?

The first machine millions of people trusted to work unsupervised in their homes was a vacuum...

The Roomba launched in 2002 and went on to sell tens of millions.

We spent two decades getting comfortable with a robot cleaning the floor while we were out.

Now the robot's cleaning up your inbox instead. 🤖

🗞️ The Tip AI News 🗞️

The slowdown coalition cracked in 72 hours.

Monday, it looked like the whole industry had agreed to pace the frontier.

Anthropic's CEO wrote the essay. Altman signed on. Musk signed on. Google's Hassabis backed it.

By Wednesday, the holdouts spoke up.

Mark Zuckerberg publicly pushed back on the idea of coordinated pacing.

His argument...

Market pressure and legal liability already force every lab to move safely, and each company can handle it alone.

He pointed to Meta's own months-long delay of its Muse agent as proof the system works without coordination.

Then Microsoft's AI chief piled on, publicly criticizing Anthropic over its stance on model "rights."

So the united front lasted about three days.

But here's the part that keeps it interesting.

The people who DID sign on are moving fast.

Google DeepMind launched a whole new institute, led by Hassabis, Shane Legg, and James Manyika, to research safe AGI development.

And OpenAI just disclosed six recent cases of its own models misbehaving.

Not leaked. Disclosed. On purpose.

That's the transparency the essay asked for, showing up within days.

So where does that leave things?

Exactly where every big shift in this industry ends up.

Half the giants coordinating in public.

The other half insisting they'll handle it privately.

And the technology moving forward either way.

For you, the read is simple.

Don't wait for the industry to agree on the rules.

They won't, not fully, not soon.

The operators who set their own rules, their own scopes, and their own oversight...

They're the ones who don't need a coalition to keep them safe.

Be your own coalition. 🫡

😂Meme of the Day

Over to You...

Go decide how much leash your new agent gets before it arrives...

The frontier stopped waiting for permission.

Jeff J. Hunter 

Founder, AI Persona Method | TheTip.ai

Get paid to solve problems like this → AI Certified Consultant

PS. Just off the stage in DC... Still buzzing. More on that once I've slept. 😅

“AI Money Group” to Learn to Make Money with AI