The morning my AI tools all quit at once

Hi {{first_name|friend}},

Something happened last week that I've been half expecting for a while and it still caught me an hour before I had to teach. Three of the big AI apps went down together and it turned into a real world test of the exact thing I keep telling folks to set up. Here's how it went.

My Take

An hour before I taught the fourth session of my Maven course last Thursday, one of my students messaged to say he couldn't do the homework. ChatGPT was down.

Then I saw why. ChatGPT, Claude and Grok were all down at once, and Microsoft Azure, the cloud service sitting under all three, was down too. A few outlets pointed at Azure as the cause. Nobody has confirmed that, but it would explain how three rivals managed to fail in the same hour.

I teach Level 1 using ChatGPT because that's what my current students use. Claude was working for me, and for a minute I thought about moving the whole session over to it. I didn't. My students had two weeks of built up context living in one too & swapping platforms in the second to last session wouldn't have been helpful.

So I did what I'd tell you to do. I checked what still worked without the app. My morning brief, a scheduled task I'd set up earlier in the week, had already run and was sitting in my Gmail. The class presentation, the course doc, my students' project files, none of that lived inside the chat window. I screenshotted the week's briefs as a backup and figured I could teach about two thirds of the session with ChatGPT completely dark.

It came back before class, slow for a while and I taught the session as written.

Here's the part worth keeping. I spent 25 years in tech and the truth is that any single system fails eventually, so you don't build your work to depend on just one. That rule didn't go away when AI appeared. A lot of us now run our day through a single app -- Thursday was a good reminder that the app is not the work. The work is your context, your documents and your decisions. The app is just where you happen to do it that day.

You can lower your own risk in about ten minutes (it's this week's action right below).

Try This Today

Back up the one thing you'd miss most if your AI app went dark tomorrow. The first file worth protecting is your About Me file, the context that makes any tool actually sound like it knows you. Don't have one yet? Build it free in about ten minutes:

Then pick anything else you'd hate to lose. A project doc, a prompt you rely on or maybe a plan you built inside a chat. Copy it into a plain Google Doc or email it to yourself so it lives outside the app. Mine is that About Me file. It sits as a plain text file on my desktop and in the cloud, and Thursday was a good reminder of why.

Watch: what Oura's IPO filing revealed about ring wearers over 50.

@aiover50

I've worn an Oura ring since 2022. Last Thursday they filed to go public, so I read the S-1 to see how typical I am. #over50 #aitechnology... See more

Worth Knowing

Washington starts arguing about AI rules

Momentum for AI regulation is building in D.C. after a prominent researcher's stark warning and lawmakers are now debating what guardrails, if any, to put in place. AI regulation and policy is still in its infancy but worth keeping an eye on.

The rules that come out of this will shape the tools we use every day. Read more

AI is changing how teams talk

Tools that draft the meeting notes and shape how a team shares information are quietly rewriting the rhythm of everyday collaboration. The tech can produce the message but it still can't read the room or know which three people actually need to hear that exact message.

Communication is where a lot of experienced professionals already shine, so knowing how AI shifts the flow lets you lead the change. Read more

California moves to ban AI that claims to read your mood

California is moving to restrict employers from using AI that says it can detect how workers feel, a claim critics call both invasive and scientifically shaky. It's a rare case of the rules catching up to a feature before it becomes normal.

Software that swears it knows your mood is exactly the kind of overclaim worth doubting. Using AI well and taking its bolder promises with a grain of salt are both important. Read more

AI is stretching healthcare into places that had none

In rural and underserved areas, AI is being used to carry the routine load so a small number of experienced clinicians can reach more people and spend their time where judgment matters.

This is the model I like to see: AI as a multiplier for experience, not a replacement for it. In the fields that matter most, we need seasoned humans to stay at the center. Read more

Tell me

Did your work actually stop last Thursday when the AIs went down or did you barely notice? Hit reply and tell me, one word is fine.