Prompting with your brain, not against it

Welcome back. If this is your first issue: hi — I'm Amanda.

August is AI for Everyday Creativity — hygiene if you use these tools at all: stay in the slow world while the products get faster. This week: prompting with your brain, not against it.

1. Add friction when mistakes are too easy.

2. Keep projects in separate places.

~2 min read

• This week: Friction before you prompt. One project, one thread, one journal page — even better, one folder (a repo).

• Try today: Sticky → one sentence out loud → new named thread + a matching bullet-journal page.

• House note: Cursor is now part of SpaceX. I cancelled. If you have an IDE or agent you trust, reply — I want recs.

• Foundation: AI Myth-Busting — What am I asking? What do I expect? How will I verify? 🔥 Fire Horse (Embrace Dynamic Energy): energy is not the same as speed. Friction protects the spark.

The tool will ship your first sentence

That is the trap.

A prompt written in the box while you are still half-feeling the idea is not a prompt. It is a blurt. ADHD brains especially: the first thing that pops out is often the loudest, not the truest. You are not bad at prompting. You are using a product that rewards the first draft of a thought as if it were a decision.

"Make this caption better." "Help me outline." "Just bounce this." Moods, not asks. The tool will answer anyway.

The 3-Question Reality Check from AI Myth-Busting still holds:

1. What am I actually asking this to do?

2. What do I expect — memory, accuracy, a finished piece?

3. How will I verify the result?

Question three is not only for facts. Caption: would I say this out loud in my own kitchen? List of ideas: which one still matters after I walk away? If you cannot answer question one in one sentence, you are not ready to prompt. You are ready for a sticky.

Friction is a feature

A pause is not you being precious. Speed is not a personality flaw — it is what these tools are for. Your job is a speed bump between the first impulse and send.

Speed bump — pick one, not five:

• Write the question on a sticky. Leave it until you can say it out loud in one sentence.

• Wait one stretch of focused work (a walk, a cup of tea, a timer you already use) before pasting anything into a chat.

• Ask Myth-Busting question one again: What am I actually asking this to do? If the answer is "help" or "idk just go," close the tab.

I still do this when the energy is high. High energy is a gift. It is also when I am most likely to dump six half-ideas into a box and call it progress. The sticky makes me pick one.

Separate the context

One long thread where you dumped the shop, the grief, the novel, and the grocery list is not "memory." It is sludge. Six unrelated midnight thoughts start answering as if they were one job. Then the output feels haunted by something you mentioned in March.

Simple rule: one project, one thread (or one folder, one agent, one workspace — whatever your tool calls it). Name it like a file, not like a vibe.

Vague (junk drawer)	Named (a context)
misc ideas lol	Shop August captions
help with writing	Zine page — river draft
creative stuff	Sourdough: weekend bake notes
work chat	Client one-pager — keep/cut list

You do not need a perfect naming system. You need a name you could find next Tuesday.

A folder is even better. The thread names the job. A folder is where the job lives — files, notes, and chat for this project only, so March's grocery list cannot answer for August's zine page. Engineers call that folder-on-GitHub a repo. You do not need to be an engineer to use a folder. Optional next step: GitHub Basics: Your Profile README as a First Project — a home for folders that can travel. No terminal required.

I keep a dedicated bullet-journal page for each new IDE chat — project name at the top, the one-sentence ask, anything I need next Tuesday without reopening the model. The page exists when the tab is closed. Same move if you never touch code.

If you already have a mega-chat: do not clean it up for three hours. Start a new one for the thing you are actually making this week. Archive the sludge. You can always reopen it.

Sticky → sentence → separate thread

A ritual small enough to do on a low-capacity day:

1. Sticky. Write the ask analog — paper, a note on the desk, a sticky on the monitor.

2. Sentence. Say it out loud in one sentence. If you cannot, you are still brainstorming. That is allowed. It is not a prompt yet.

3. Separate thread + journal page. Open a new chat named for this project. Start a dedicated bullet-journal page for that chat. Paste the sentence. Then — and only then — add the extra context that belongs to this job.

That is prompting with your brain. The model can still be fast. You do not have to be.

Community Corner

Question I get: "If I start a new thread, won't I lose the context?"

My answer: You will lose the sludge. That is the point.

The context worth keeping is the context for this project — the caption series, the recipe test, the one client page. A mega-chat feels like a second brain because it is full. Full is not the same as useful. Name the project. Start clean. If you need an old detail, go get it on purpose instead of letting March answer for August.

If a tool cannot start a new thread without dragging the whole history along, that is information about the tool. Keep less in it — or skip it today.

Question I have for you: What are you using instead of Cursor — another IDE, another agent, or you walked away this week? Reply. I am collecting recs, not running a purity test.

Personal update

Cursor is officially part of SpaceX as of this week — Elon Musk's company. That is enough for me. I am also not willing to fund the data-center buildout those models run on.

I cancelled my subscription. I am starting to try other IDEs and agent tools. I do not have a replacement to bless yet.

If you came here through the May workbench posts: the habit still holds — a folder as a project home, named context, an analog note. The vendor is what changed. GitHub Basics is still the safe first project.

If you have something you trust, reply and tell me what — and why it still feels like yours. Staying on Cursor does not make you the enemy of this letter.

Your tech struggles, reflected back. Got one? Send it in.

Think about the last time you opened a chat before you knew the ask:

1. Where did the first impulse go — sticky, notes app, or the box?

2. How many unrelated projects are currently living in one thread?

3. Can you say this week's ask in one sentence? If not, what are you actually trying to feel out?

4. What would change if the tool had to wait for you?

Hit reply. I read every one.

🔥 The Fire Horse's Callout: What to Charge Forward With

Embrace Dynamic Energy — then aim it. Fire that goes everywhere is just burn.

One sticky. One new thread. One project name.

What's next

Next Friday: chat vs desk vs actually making the thing. Typing the first thought into a window is a conversation. Leaving a note on the desk is how an idea stays real. Neither one is the project yet.

— Amanda

Resources

Resource	Best for
AI Myth-Busting Toolkit	3-question reality check before you prompt
GitHub Basics: Your Profile README	A project folder that travels (repo = folder + history)
Free resources hub	Guides, no paywall

• Subscribe to the newsletter → pythonessprogrammer.com/newsletter/subscribe

• Browse free resources → pythonessprogrammer.com/resources

• Shop → stickyspells.etsy.com

• Support → pythonessprogrammer.com/support

• Forward this issue to someone whose mega-chat has become a second job

Philosophize your tech use. Tech that works, for people like us.