How I Used AI to Build My Weekly Business Operating System

What happens when someone who doesn't think in systems finally builds one

Nov 18, 2025

It was Friday morning, it was raining, quiet, and I had no calls. (Time lottery! Even though I love my calls & feel energized when I’m done, there’s something about a wide-open day to myself that feels luxurious.)

I was sitting with my coffee, and I realized: I have no idea what I’m supposed to be doing next week. 🤪

Not in a “I have nothing to do” way. The opposite. I have a client to serve, a newsletter to write (this week will be #250 by the way - 250 consecutive Thursdays 🥳), videos to record, a quiz to build, and a business to run. But when I looked at my week ahead, I just felt... scattered (i.e., overwhelmed).

You know that feeling where you’re busy but you can’t quite articulate what you’re busy WITH? That was me.

So I did what I’ve learned to do when I feel like this: I opened Claude and said, “I need help.”

And what happened over the next few hours (followed by a handful of additional sessions with Claude) was a great reminder that AI can really help me buy back my time.

My brain doesn’t naturally think in systems.

I get that that might seem contrary, considering I’m such a fan of AI, but it’s actually AI that has changed how I think about systems. Because I also use AI to teach/guide/ mentor me when it comes to systems (think the technical side of systems, such as creating a Notion CSV file that I can then import and iterate on), my brain is adapting to a different way of thinking.

It’s not so much that I’ve resisted structure, but rather that I wanted a gentle one. It’s like the petulant child in me stomped her foot and said, “I don’t work for myself to 'have to' do things. I enjoy creating, iterating, and adjusting on the fly.”

However, at this stage in my life, I know that not having better structure (i.e., better systems) won’t get me where I want to be.

AI has become a consistent tutor.

Showing me ways to use Notion I’d never thought of. Creating structures I wouldn’t have known to ask for. Building systems based on how I actually work, not how people who love productivity systems tell me I should work.

So now? I’m building new systems. But they’re MY systems - flexible, iterative, designed around my reality.

Because at this age, I believe that the best systems aren’t the ones we force ourselves into. They’re the ones you build that help you see patterns you can’t see on your own, and show you possibilities you didn’t know existed.

The Conversation That Changed Everything

So, back to that Friday morning. I opened Claude (my AI tool of choice for strategic work - more on why in a minute) and I started talking.

Not prompting. Not commanding. Talking (WisprFlow has saved my life, lol. I have officially hit over 600k words dictated!):

“I feel scattered. My weeks are getting busy, and I don’t know what to prioritize. I need to regroup and recap what we’ve been working on, what my priorities are, and create an action plan with a manageable system moving forward.”

That was it—no perfect prompt. No frameworks. Here’s my problem, now let’s solve it.

And Claude responded with: “Let me pull up what we’ve been working on and create a comprehensive recap.”

Wait, what? 🤯
(That was for effect, lol, I knew it could do that.)

I’m working in what’s called a Claude Project - which is basically a workspace where Claude remembers all our past conversations. So it could see:

The new quiz we’d been planning
Client work we’d discussed
Apps we’ve discussed (including learning and understanding database structure, context, and code constraints)
My focused business strategy
Content strategies we’d mapped
My newsletter history
My personal life constraints (caregiving for dad, energy limitations, upcoming holidays)

This created version 1.

Between version 1 (plans for what to create over the weekend) and when I returned to Claude, I had another trip to the ER with my 81-year-old father (he has heart issues, so this wasn’t a huge shock - and he’s doing well, but it certainly threw the weekend for a loop).

Which ended up being the break I needed between version 1 and the next iteration of the conversation.

This Is Where It Gets Good (The Part About Not Taking First Outputs)

Too often, people ask AI a question, get one response, and think, “Okay, that’s the answer.”

As an example, they might ask ChatGPT for:

A content calendar, get a response, and try to force themselves to follow it.
A weekly plan, get something that looks professional, and wonder why they can’t stick to it.

The problem isn’t the AI. The problem is treating it like a magic 8-ball instead of a thinking partner. 🎱 (first time I’ve used that emoji, lol)

So, when Claude gave me that first comprehensive recap, I didn’t just say, “Great, thanks!” and walk away.

I came back the next day(after picking up my dad from the hospital, actually) and said:

“This is good, but it’s not quite right. The world-building app? Not a priority right now. I want to test with a custom GPT first. And I need this to account for the fact that Thanksgiving is next week and dad’s health is unpredictable.”

And Claude said (I’m paraphrasing): “Okay, let’s revise everything based on your actual reality.”

It created an entirely new document with different priorities, structure, and expectations.

Then I came back again and said: “This weekly rhythm is still too rigid. I have calls scheduled, and I’m working in more than 90-minute blocks (Claude was breaking things into 90-minute blocks because of the head injury I had almost two months ago).

Third revision. Even better.

This is the part people miss: The gold isn’t in the first output. It’s in the iteration.

Why This Isn’t Just “Using ChatGPT”

Quick sidebar, because I know someone’s thinking: “Can’t you do this with ChatGPT?”

Sure. I use both (and Gemini). My preference, in order, is Claude, then Gemini, then ChatGPT. Obviously, the more you use one particular LLM, the better it should be for your brand, but Claude tends to be my “go-to” most of the time.

The big one: I’m building my “second brain” in Cursor.

My brilliant friend Jason Resnick got me into Cursor (which, by the way, Claude will of course help me master so I don’t drive Jason crazy, lol) and got me started on the second brain process he created.

I can send things directly from Claude to Cursor. All these markdown files we created? All this context? It goes straight into my knowledge base. That integration alone is worth it for me.

I’ll explain the difference between a Claude project and the Cursor second brain in another post.
(ChatGPT would require a workflow to accomplish this. Claude gives me a button 😉. The screenshot below is from within my Claude project).

The context thing:

I’ve been using Claude more for most of my projects, which means there’s just more available to access. More conversations about my projects. More understanding of my constraints. More history of what we’ve built together.

So, when I sat down on Friday morning and said, “I need help,” Claude could draw on weeks of past conversations. It already knew about everything I mentioned above (quiz I’m rebuilding, apps I’m starting over with, newsletter, etc.).

What We Actually Built (The Tangible Stuff)

Over those two hours on Friday morning, then broken up by hospital trips, coffee refills, a couple more sessions and just letting my brain rest - Claude and I built:

1. A comprehensive business recap showing every project, every priority, every constraint (8,000 words, downloadable markdown file)

2. A revised priorities document based on my ACTUAL life - caregiving responsibilities, holiday season, energy levels (not fantasy planning, ya’ girl isn’t going to hustle)

3. A flexible weekly rhythm guide - not “Monday 9 am do this” but “here are your Monday priorities”

4. A complete Notion database structure to track my weeks - not just plan them, but see patterns in what I actually DO versus what I THINK I should do (I’m going to utilize Notion’s AI to recap the week, tell me what I missed, what got done, and use that for moving things into the next week or making adjustments).

5. The actual CSV file ready to import into Notion - not instructions for me to build it myself, the actual file.

However, the Notion file also took a few iterations. The first version had me scrolling to the right for days. It was a little ridiculous and not remotely efficient.

I don’t know the actual time investment for this entire project, but if I had to guess, I’d say it was probably a full day’s worth of work. If I were to have done this manually? It’s safe to say it would have taken me multiple days of work at least (I have no doubt my frustration level would have derailed my progress).

The System I Built (For Someone Who Doesn’t Think in Systems)

The centerpiece of all this is what we’re calling the Weekly Rhythm Database (cue Janet Jackson… ”We are a part of the Rhythm Nation” 🎶... couldn’t resist. Thank you 1989).

It’s a Notion database where I track daily tasks each week. Not just what I plan to do, but what I actually DO.

Every task, deliverable, and reflection (reflection and analysis will be done on Friday).

What makes it work for me is that there are checkboxes for:

Schedule social posts
Write newsletter
Client calls
Video creation
All the usual stuff

(Who doesn’t like checking things off their list?)

But there are also fields for:

Energy level (because I’m human - and life has a way of sneaking up on you. Best laid plans and all)
What worked (so I can do more of it)
What didn’t work (so I can adjust plans moving forward - this is a judgment-free zone, it’s all about learning what works)
Life notes (dad’s health, holidays, whatever is happening)
Week rating (was it a good week? a survival week? somewhere in between? I chose “good, better, best” - hat tip to a friend for sharing that with me)

After a month of tracking, I’ll be able to ask Notion AI: “What tasks do I consistently complete versus consistently skip? What does that tell me about my actual priorities?”

That’s data I can use.

Not guilt. Not “I should be better at this.” Actual information about how I work.

AND… I have a feeling that after I use this for a few months, I’ll end up iterating and refining it as well.

What This Means for You (The Actual Takeaway)

You don’t need to be a “systems person” to build systems.

You don’t need perfect prompts to get value from AI. Instead, look at the quality of your outputs. I consider prompting a skill that requires mastery - it’s an ongoing improvement process, not a one-and-done.

You don’t need to take the first output and force yourself to fit into it.

What you need:

A real problem (feeling scattered counts)
An AI tool that can handle context (Claude, ChatGPT, Gemini, whatever)
Willingness to iterate (”that’s not quite right, here’s why...”)
Trust to build something that fits YOUR life

The framework is simple:

Start with your problem. Not “I need a content calendar.” But “I feel overwhelmed and don’t know what to prioritize.”

Have a conversation. Don’t try to craft the perfect prompt. Just explain your situation. Your constraints, what’s going on in your life, all of it. Ideally, ask AI to ask you questions about what’s going on. It’s hard to create a plan if you can’t define what needs to be included.

Iterate. When the first output isn’t quite right, say so. “This is good, but it doesn’t account for...” Claude, Gemini, or ChatGPT will adjust (or whatever tool you’re using).

Get tangible outputs. Ask for files. “Can you create this as a CSV I can import?” “Can I download this as a markdown file?” Build assets, not just chat history (Claude can also create JSON files for n8n and Make automations)

Use it in your actual life. The system only matters if you actually use it. And you’ll only use it if it fits your reality.

*BONUS TIP*
This process was so helpful to me that, of course, it made sense to turn it into content (#everythingiscontent). But not just a “here’s what I did” post (like this one). I’ve got a video and the Notion template I’ll be sharing with you, too (soon!).

The Part About Newsletter #250

I mentioned I’m publishing my 250th consecutive newsletter this week. That’s almost five years of showing up every Thursday without missing a week (there aren’t 250 on Substack. It went from Kit to beehiiv to Kit, and now Substack. But that’s a conversation for another day, and it was originally called #FtheHUSTLE).

It’s crazy when I think about what’s transpired in the last five years.

The simplest reason I’ve stuck with it for so long?

I enjoy it. 🥰

Never underestimate the things that light you up.

What’s Next (If You Want to Try This)

I’m going to do something a little different with this.

I’m making all the files available - the comprehensive recap template, the weekly rhythm guide, the Notion database CSV, all of it.

Not as a paid product. Just as: here’s what’s possible when you stop thinking of AI as “write my caption” and start thinking of it as “help me build systems for my business.”

Next week, I’ll show you the Notion setup - how to import the database, how I’m using it, and what I’m learning after the first week of real data.

But today’s article is about this: Stop taking the first output from AI. Start having conversations. Start iterating. Start building systems that fit YOUR actual life.

Even if you’re not a “systems person.”

Especially if you’re not a systems person.