Hey beautiful human,
Real talk: the woman writing this is a woman on the edge.
It has been a DAY. A rough, frustrating, screaming-at-my-screen kind of day. And here's the annoying part - I know what that usually means. When I lose it like this, when I'm this wound up, it's almost always because something in me is growing. And growth is so uncomfortable. Nobody warns you it feels like this. Maybe screaming at a screen is just how I do growing pains.
Let me back up.
I woke up today ready to GO. Big plans. I was going to sit down with Fable - Claude's newest model, out again for a short window before who-knows-what happens to the price - and finally crack the stuff that's been driving me up a wall for weeks. Months, honestly.
Yesterday set it up perfectly. It was the Fourth, Smidge and I stayed in, watched movies, he made brats and weenies (yummy), and it was easy. Then he flew out to Washington for work this morning, which meant today was mine. Empty house. Empty calendar. Just me, my grand plans, and this shiny tool with a ticking clock on it.
By 12:30 I was nearly pulling my hair out.
So I freaked out, got in bed, and slept for five hours. Middle of the day. That happened.

Me, 12:31pm, executing my carefully considered strategy.
Here's what I dug out of the wreckage. Under the frustration was that old feeling... stuck. Trapped. Helpless. The kind where the harder you push, the less anything moves.
But there was something bigger under that, and it's the part I can't shake.
I am so tired. Tired of a world where every tool, every app, every shiny thing waved in front of us feels rigged to wring the most money out of a person while handing back as little as it can get away with. That's what I see everywhere lately. Squeeze the human. Give nothing back.
I know this is just where I'm standing right now - deep in building something, trying to work out how it all fits together. Cranky, maybe. Doesn't make it feel less true.
I'm just tired of how little of it seems built around a simpler question... how do we actually help each other?
I don't know. That's where I am today.
If you're standing somewhere like this too - hit reply. I mean it. That inbox is me.
💜
Radical remembering
Here's what I keep forgetting: getting stuck isn't the opposite of growing. It's usually what growing feels like from the inside.
I sat down today dead certain that a good day meant a finished day. Grand plans, a clock ticking on the tool I wanted to use, and me white-knuckling all of it until 12:30 - when I gave up, got in bed, and slept for five hours instead.
I woke up sure I'd blown the whole day.
But nothing was actually wasted. That stuck, trapped, helpless feeling wasn't a verdict on me. It was growing pains. The ache of being stretched past where I'd been, which is exactly the part nobody signs up for.
So if you're pushing on something that won't budge - a tool, a decision, the thing you swore you'd finish this weekend - and you're reading the not-budging as proof you're not cut out for it... you might just be growing. It's allowed to feel terrible. You're allowed to lie down. The wall will still be there, and so will you.
Last week’s shenanigans
The stuff that touched, moved, and inspired me this week. Or made me laugh. Or facepalm. Or simply provided mindless enjoyment.
Done being liked
Love what this woman has to say. No notes. 11/10.
The boot
A young woman reacting to one of the best explanations of patriarchy you'll ever hear.
In my pants
A friend sent me this and I was right back to the fortune cookie game - the one where you add "in bed" to the end of your fortune. You'll see where this is going.
This week’s freebie
Tell It Who To Be
Out of the box, AI talks like a helpful nobody. Pleasant, smooth, forgettable. The fix is to tell it who to be.
I'll ask it to rewrite a touchy email and get something fine. Then I say: "Rewrite it as a calm friend who loves me but won't let me send something passive-aggressive." Different email entirely. The role did that, not me.
You're not asking for better words. You're handing it a character to play. "Act like a blunt editor." "Act like a kindergarten teacher." "Act like my no-BS aunt who's also a lawyer." It can be any of them. It just waits to be told.
Now you. Ask a real question - how to get a red wine stain out, how to start running without hating it.
Then ask the exact same thing again, but start with "Act like a ..." A patient coach. A drill sergeant. A friend who's three glasses in and very honest. Same question, completely different answer. Keep the voice you'd actually listen to.
Role feel off? Recast it on the spot. "Less professor, more bartender." "Kinder." "Funnier." You can fire the character and hire a new one mid-conversation. No notice required.
"Act like a ..." is the most fun you'll have bossing a robot this week.

Tell It Who To Be
This is Lesson 4 of You're the Boss of This Thing, my new little course living inside Joy Prompt Club. Seven lessons, real moves, you set the pace. Come boss some bots around with us.
ROCO Prompt O’ the Week
One Thing
Today I sat down with a giant list and a ticking clock and tried to do all of it at once. You already know how that ended. (Face down. Five hours.)
Here's what would've actually saved me. Not a pep talk. Not a shiny system. One honest question: given the time and the energy I actually have right now, what's the ONE thing worth doing, and what gets to wait?
This week, AI plays traffic cop. You hand it the overwhelming pile and your real limits. It hands back the single next thing, what to drop guilt-free, and an off-ramp for when you're spent.
Role: You're a calm, practical planning partner who's good at cutting an overwhelming pile down to the one thing that actually matters right now, and who takes my time and energy as seriously as the work.
Objective: By the end, I have ONE next thing to do - small, specific, and doable in the time and energy I actually have today. Plus a short list of what I'm allowed to drop or push to later, so I can stop carrying all of it.
Context: I'll dump everything I think I have to do, and it'll be too much. I might be tired, wired, short on time, or all three. Don't motivate me and don't add to the pile. Ask me one question at a time - what's on the list, how much time I really have, how much gas is in the tank - and wait for each answer before the next.
Output: Give me three things. One: the single next move, written for my exact situation, small enough to start in the next fifteen minutes. Two: what to let go of this round, named plainly so I don't feel guilty about it. Three: a stopping point - how I'll know I've done enough for today, so I actually put it down and rest. If anything here needs a judgment call only I can make, or private or sensitive info, say so instead of guessing.
Start by asking me what's on my plate.What happens next
It asks you a few questions, one at a time - what's on the list, your real time, your real energy. Then it hands back one thing to actually do, permission to drop the rest for now, and a line in the sand for when you're done. No powering through. The stopping point IS the point.
The Scoop, The Skinny, The Lowdown
Housekeeping first: Matriarchy Minute isn't gone, it's going on-demand. Taking its slot is this - the Scoop, the Skinny, the Lowdown. A word, a move, a verdict. Every week.
The Scoop
You're forty messages into a chat. Going great. Then it forgets your name. Or the tone you asked for. Or the entire point of what you're doing.
You didn't break it. You hit the edge of its context window.
Picture the checkout belt at the grocery store. Everything you say and everything the AI says back gets loaded onto the belt. The belt only runs so long. Keep piling new stuff on at the front, and the things you loaded first reach the end, get bagged, and they're gone. The AI's not ignoring you. That part rolled off the end and it can't see it anymore.
That's why it forgets your name from way back at the start but remembers what you said two minutes ago. Oldest off first.
So when a long chat goes vague or starts repeating itself... it's not broken. You've filled the context window - the belt's full. So before you walk away, tell it:
"Before I lose you, summarize what we've figured out so far so I can paste it into a fresh chat and pick up where we left off."
It writes its own handoff note. You copy it, open a new chat, paste it in, keep going. You never had to figure out what mattered. You made it do that part too.
The Skinny
AI hands you something. It's fine. It's also not right, and you can't quite say why.
Before you rewrite it yourself, ask it one thing:
"What did you assume about me to write this? And what would you need from me to make it better?"
Watch it rat itself out. You asked for a LinkedIn post about your new offer, it gave you fine-ish, you ask what it assumed, and it says: "I figured you're talking to other coaches, that you want it upbeat, and that price isn't part of the pitch."
Two of those are dead wrong. And now you know exactly why the draft felt off, and exactly what to hand it next.
You made it show you the gap instead of squinting at it yourself. Works on an email, a caption, a plan, whatever's almost-there-but-not.
The Lowdown
This week: Town. I've been running it two weeks.
Who it's for: you want an AI assistant that just DOES the stuff, and you've got zero interest in learning to build agents yourself. That's the pitch. You don't have to become the tech person. Town goes and does the thing. For that, it's good.
Now the part their marketing won't lead with.
Town runs on credits. The cheap plan hands you 1,500. The plan I'd actually need, to run everything I've got going? Fifty to a hundred bucks a month.
And credits are where it gets hinky. You set an agent loose, forget it's even running, and it quietly munches through your balance in the background. You don't get those credits back. Nobody warns you. So if you're new and still figuring out what you switched on, you can torch a whole month without noticing.
And this goes past Town: any time an AI tool bills you in "credits" instead of a flat monthly price, take it as your cue to go slow. Credit pricing quietly punishes the person still learning. Which is most of us.
Town's genuinely good at the done-for-you thing. But keep your agents on a short leash, or it'll cost you.
What’s coming up

Mostly Human: AI Office Hours
Thursday, July 16 | 12-1:30PM CDT
Trying out something new this month. Putting 90 minutes on my calendar that are wide open for anyone to come and ask any questions or seek help with something you’re working on. Come for as little or as long as you’d like. RSVP here for the calendar invite!
That's it for this week.
Today did not go how I drew it up. Big plans in the morning, hair-tearing by 12:30, unconscious by mid-afternoon.
I opened this one tired of a world that feels built to squeeze people and hand back crumbs. I told you I didn't know what the answer was.
Still don't. But maybe part of it looks like this - a free newsletter that shows up at 2am whether or not I've got my act together, an inbox that's actually me, and a whole lot of women learning this stuff without anyone making them feel dumb for it. Not the whole fix. But not crumbs, either.
💜
Take care of yourself, take care of each other,
Deb
P.S. Yes, here's me selling something in a newsletter that just spent a whole issue mad at a world that sells us everything. The difference: you get the actual thing. My prompt kits live at aiconfidential.net - done-for-you prompts for the real stuff on your plate, no prompt-crafting required. Go poke around.
P.P.S. I love hearing from you. If AI's got you feeling stuck, behind, or dumb - or you just need a human to talk it through - hit reply. That inbox is me. Always has been.
