Hey beautiful human,
Emily and I have a campaign in flight, and it turns out a campaign is forty jobs sharing one name.
Most of the forty are lists.
Cross-check this one against that one. Confirm who's on which. Open it again because what was true on the last pass has wandered off since.
I'm good at that. Give me four spreadsheets and I'll find you the one row that's lying.
Then I need a lie-down and a juice box. Ask me a simple question at 3pm and watch me blink at you like a startled owl.
Me, 3 pm, on the receiving end of a yes-or-no question
Gave the list work to AI. Fine. Good. Correct decision.
Then didn't check it.
Two people had to ask me twice to come off a list. The second ask had a tone. Deserved tone.
I'd taken them off the spreadsheet. The form feeds the spreadsheet. The form still had them.
Brand new connection I'd never used before, thirty-nine other things open, and it occurred to me to ask about the connection but then I quickly lost the plot.
Cripes.
Unsubscribes used to land on me like a personal rejection. Now they read like calibration... somebody raising a hand to say this isn't for me, which is information, and a kindness, and none of my business.
Two irritated humans is a different animal. That one I earned.
I learn by doing things wrong in front of people. Cannot be told. Have tried.
On a mission to fail faster now. Faster, and preferably in smaller pieces.
I've been teaching this stuff on one assumption. Fear. Get a woman past scared and she's off to the races.
Fear's in there. It's smaller than the thing standing behind it.
<< Scene: any Zoom call, anywhere >>
MAN: Yeah, I had AI do the first draft.
ROOM: Smart. Good use of the tools.
WOMAN: Yeah, I had AI do the first draft.
ROOM: ...so did YOU write it, or.
<< scene fades >>
Women are learning this stuff everywhere. In DMs. In group chats. In little rooms with the door shut, teaching each other what actually works.
Talkin' Bout a Revolution 🎶
Tracy told us how this one starts. Quiet. Low. Down where nobody important is bothering to listen.
I love those rooms. I run one.
And I want us louder now.
Whatever gets written in public right now is what these models read next. That's the actual mechanism. The people who've had the microphone on this technology have mostly been men, and the machine has been taking notes the whole time.
If we want it to know what we think, we have to put it somewhere it can see us.
I know what I'm asking, by the way.
In rooms where I'm one of the few women, I say a thing and it comes back to me explained. Slowly. By someone who heard the sentence and decided I couldn't have meant it. Then somebody argues I'm wrong, then somebody else joins in, and it is not only the men doing it.
I do it anyway. It costs, and I've got no fix for the cost, which is annoying, because I like my answers to fit on a slide.
F***ing furious about it, y'all. Also stumped. Apparently those two share a room.
Elsewhere: a course from Abi Awomuso that blew my doors clean off. Every time with that woman.
And I signed a part-time PM contract with a local seasonal event marketing team, through December. More money than that ill-fated June contract for about a third of the hours. Covers the utilities without me doing sums at 11pm.
Have you ever left the AI part out when somebody asked how you did it? Or had to learn something the hard way that you already knew perfectly well?
Radical Remembering
Here's what I keep forgetting: nobody is transcribing my DMs.
Some of the best teaching I do happens with the door shut. A group chat. A live session. One woman, one question, exactly the thing she needed at the exact moment she needed it... and then it's gone.
That work is real. It changed her afternoon.
It also lands nowhere the machine will ever find it.
It feels like output. It isn't stored anywhere.
There's a whole network of us out here. Brilliant. Generous. Entirely off the record.
The small room is where I get brave. Somebody still has to walk out of it and write the thing down.
Last Week’s Shenanigans
Four again this week. If you're new here, this is where I share the
stuffs I encountered on the interwebs that touch, move, inspire, or
simply make me snort-laugh.
The hardest truth
What's the hardest truth you've had to accept? That's the whole question. Give it five minutes and see what your brain coughs up before you've had a chance to make it presentable. 4:43 duration.
Blindspots begone
Stanford came up with a prompting method that keeps AI research from going narrow and biased on you. Gonna take this puppy for a spin. 1:11 duration.
Stop wifing-up your job
Neurodivergent people burning out trying to fix somebody else's problem, because we were taught we have to fight to take up space and fight for justice inside systems built to be unjust. Her advice is to treat the job like the ho it is and stay committed to yourself first. 1:45 duration.
Like hootenanny but better
Snort-guffaw-groan. Which, for the record, is the exact reaction Smidge works very hard to produce. 10 seconds.
Prompt O’ the Week
Where Else Does This Live
Up in the opener I told you about the two people who had to ask me twice.
I deleted them from the spreadsheet. I watched the rows disappear. Job done.
They kept turning up.
The spreadsheet wasn't where they lived. A form put them there. The form still had them, and every time it ran, it politely put them back.
I had been editing the copy. The copy gets refreshed from the source. The source always wins.
That's the whole lesson, and it applies to nearly everything you'll touch this year.
Your contacts sync from somewhere. Your calendar pulls from somewhere. That project tracker didn't type itself. When AI hands you a tidy sheet of anything, it built that sheet FROM something, and that something is still sitting there with the original in it.
So the question isn't "did I delete it."
The question is where else does this live, and which copy is upstream.
Now, the honest part.
AI cannot see your stuff. It doesn't know your form feeds your sheet. Ask it where your data lives and it will invent an answer that sounds fantastic and is entirely fictional.
What it CAN do is interview you. It knows how these tools tend to chain together, so it plays the person who's set this up a hundred times and asks the questions you didn't think to ask.
You barely type anything. It does the work.
ROLE (who AI should be)
You're the person who has set up this exact kind of tool connection a hundred times and has watched everybody make the same mistake.
OBJECTIVE (what good looks like)
Interview me until I know every place this information lives. Success is that a person asks to be removed ONE time and is actually gone. A neat list I never verify is a failure.
CONTEXT (what you know and don't)
I'm changing or deleting something. [What, in one line.] Tools I think are involved: [list them, guessing is fine.] I do not know how they're connected. That's why I'm here.
OUTPUT (the shape of the answer)
One question at a time, wait for my answer, plain words.
Start with how the information got in there in the first place. The second copy is almost always upstream of where I noticed.
Ask who else touches this. People, not just tools. Exports, downloads, shared files.
If I say "I don't know," don't move on. Give me one sentence on where to look, and wait.
If I name a tool you don't actually know, say so instead of guessing.
Stop when I've gone two questions without naming a new place.
Then: the list, source first, copies after. Never a copy before the source.
Last line, every time: tell me to go open each one and confirm with my own eyes, and come back and tell you what I found.The Scoop, The Skinny, The Lowdown
Three quick hits, every week. The Scoop: one AI word, decoded. The Skinny: one move you can steal today. The Lowdown: one tool, one honest verdict, from my own use. A word, a move, a verdict. In and out.
THE SCOOP
GPT
Everybody types it. Nobody says it. It stands for Generative Pre-trained Transformer, which is a spec sheet somebody accidentally started using as a name.
Generative: it makes new sentences instead of going and fetching one that already exists. Writing, not searching.
Pre-trained: it did all its reading before it ever met you. Everything it knows came from what people had already written down and put somewhere it could reach.
Transformer: the engineering. You will never once need this.
The middle letter is the one to keep, and it's the thing I was banging on about up top. Pre-trained on what got written down. Whatever ended up in there is whatever somebody bothered to publish.
THE SKINNY
Tell it who's reading
"Write me an email about the schedule change" gets you an email for nobody. Bland. Fine. Instantly forgettable.
Try this instead: "The person reading this is 58, has been here twenty years, hates being managed, and will read exactly one paragraph before deciding how she feels about me."
Same request. Completely different email.
Most of us describe the TASK and skip the human. The human is the part that changes the words.
Works on anything. The reader is my sister who thinks all of this is a scam. The reader is a hiring manager with ninety seconds and forty other applications. The reader is my mother and she worries.
Describe them accurately, warts and all. The AI isn't going to tell them.
THE LOWDOWN
This week: Copilot
Everybody thinks Copilot is just ChatGPT wearing a Microsoft badge. It was close to that once. Not anymore.
Microsoft builds and runs it. OpenAI's models do a lot of the work under the hood, and these days so do Anthropic's Claude and Microsoft's own homegrown models. You don't pick. It picks.
What it's genuinely best at: living inside your Microsoft stuff. It can see your actual emails, your actual files, your actual meetings, and answer from them. Nothing else does that with so little setup. If you work in Outlook, Teams, Word and Excel all day, this is the least amount of friction you'll find anywhere.
So here's my line on it. If you use AI for work and only for work, and your work lives inside Microsoft, stick with Copilot. You're fine. You're better than fine.
The second you step outside Microsoft, or you want to make something, you'll start hitting walls. Connections to non-Microsoft tools are shallow. And it's tightly guardrailed on creative work in a way ChatGPT and Claude aren't - it'll decline things they'd cheerfully hand you.
One more catch: the free version is a completely different animal. Free Copilot searches the web and chats. It cannot see your files. The whole reason to use it is behind the paywall.
Cost: free to chat. Individual paid sits around $10 to $20 a month depending on which bundle you land in, and Copilot is now baked into regular Microsoft 365 Personal and Family whether you wanted it or not. Work licenses run roughly $18 to $30 per person per month on top of a base Microsoft license. Prices moved twice in the last year, so check before you buy.
Verdict: the best tool in the building if Microsoft IS your building. Frustrating the moment you want to work anywhere else.
What’s Coming Up
Summer School with Em & Deb: Session 2
Make Your Resume + LinkedIn Easy to Understand and Hard to Ignore
Tuesday, August 11, 2026 | 11AM CT/12PM ET
Last week we figured what employers want (job descriptions give you the clues) so this week we’re gonna talk about positioning, proof, and packaging. Emily & I run it live with the replay and materials available to all registered students.
Em & Deb: LIVE on LinkedIn
Thursday, August 13, 2026 | 11AM CT/12PM ET (30 minutes)
This Thursday it's Deb's turn to drive: how to use AI to work FASTER on your search - resumes, outreach, research - without everything coming out bland. Remember: AI is an overenthusiastic intern. Tireless, eager, and absolutely not to be left unsupervised.
Prompt Along: AI Improv
Tuesday, August 18, 2026 | 7PM-8PM CT (Complimentary)
Prompt Along: AI Improv is my free monthly hour where we play with AI out loud, together. You bring the questions. You bring the curiosity. You bring that one tough problem you've been poking at for three weeks. And we prompt our way through it, live.
WDAI Learning Lab
Prompt Along: AI Improv
Thursday, August 20, 2026 | 1-2PM CT (Membership required)
The one rule of improv is "yes, and." That's the vibe. You throw something out, we run with it, we watch what the AI does, we laugh when it gets weird, and you leave with something you can actually use.
Bring one of these:
• A question you've been a little afraid to ask
• A task you keep avoiding
• That problem living rent-free in your head
That’s it for this week.
I'm going to go fail at something small next week.
Nothing with a client's name on it. Something where the worst case is I go "huh," and then I go again.
Any takers? Tell me what you broke. Those are my favorite emails.
And if you read all that up top and thought "not in my building" - I believe you. You know your rooms and I don't. There's no prize for volunteering in a room that's going to charge you for it.
I'll keep saying it out loud. Happy to go first for a while.
💜
Take care of yourself, take care of each other,
Deb