CustomGPTs for Substack Creators

Why, when and how to build GPT's to level up your Substack

Author’s note: This is a cross post from a collaborative article that was originally published in Jose Antonio Morales`s newsletter, Automato and is written from his perspective.

The 25-hour monthly tax

Understanding how to get the most out of ChatGPT isn’t obvious at first. What is obvious, once you see it, is how much time disappears into repeated chats.

Timo made it very obvious to me:

You waste 25 hours a month and don’t even know it.

Every time you open ChatGPT to write a Substack Note, draft an article, or repurpose content into a different format, you run the same exhausting workflow:

Put in your basic prompt, re-explain your tone, add context the AI forgot, wait for mediocre output, then edit for 10 minutes so it sounds halfway like you.

10 minutes per prompt. 5 prompts a day. 50 minutes daily.

If my math is correct, that’s 25 hours a month babysitting ChatGPT instead of moving your personal brand forward.

Just imagine what this means for a team. But as Timo suggests, this is not only a money-saving situation; it is about how to make your work better, easier, and make the best of your investment with AI.

---

Custom GPT’s: Your Personal Automation System

Without having to go too deep into AI and all the possibilities it gives, in the case of ChatGPT, you have the option of creating Custom GPTs, which are specialized versions of the traditional ChatGPT, trained by you, with special instructions you provide in advance.

Custom GPTs are great because you will never have to give the same instructions at every new chat you start.

Custom GPTs go beyond “better prompts,” they automate your repetitive tasks.

You train them once with your instructions, tone, and examples. They store everything. You paste your input and get results.

Multiply that across every repetitive content workflow, for example:

✓ Writing hooks (10–15 minutes)

✓ Email subject lines (5–10 minutes)

✓ Generating content ideas (10–15 minutes)

✓ Repurposing Articles Into Notes (15–20 minutes)

When Timo talks about a personal automation system, this is what he means: identifying recurring tasks and letting Custom GPTs handle them.

Finding those tasks isn’t always obvious at first. ChatGPT isn’t a solution to everything. But once you identify what canbe automated…

…it works like a charm.

---

Slowing Down (On Purpose)

Timo’s instinct identified that enthusiasm as a potential risk, and I believe he is very right:

So the question is, why aren’t you using the best out of Custom GPTs?

How much longer will you spend your best hours on repetitive tasks?

Okay, all fine, but wait, wait, wait,...

I know you’re already planning a whole GPT content battalion inside your mind, but most creators mess up right here.

They try to automate every aspect of their creator business, get overwhelmed by all the possibilities, and give up before the CustomGPT’s workflows are perfected.

I came up with a smarter way to decide what to automate and when.

Let me show you the exact decision framework I use to know when to build a Custom GPT (and when it’s not worth the effort yet).

A three-step process to lower the speed

The thing is, how to define what is automatable with Custom GPTs?

Timo came up with a simple process:

Not every task needs a Custom GPT.

Before you automate everything in sight, run it through this simple filter. You’ll dodge building systems you don’t need and skip automation that doesn’t actually help.

Filter #1: The Only Once Rule

If you only do this task once, skip the Custom GPT.

Example: You write the “about you” page of your newsletter. You’ll probably never write that type of story again. Just use regular ChatGPT and move on.

Save Custom GPTs for tasks you repeat weekly and daily.

Filter #2: The Human Factor Rule

Some tasks require human judgment, emotional intelligence, or ethical consideration that AI can’t replicate. I wrote an article with Sam Illingworth about it.

If I cut it down to one single conclusion, it’s: don’t automate human relationships.

Filter #3: The “I Want To Do It” Rule

If you love doing the task, don’t automate it.

Example: Maybe you love brainstorming content ideas or drafting first versions of articles by hand. Great, keep doing that. But if you hate repurposing that article into 10 different social posts? Build a CustomGPT for it.

Automate what drains you, keep what energizes you.

The Decision Framework (In One Line)

Your one-sentence phrase to remember:

If you do it more than once, AI can ethically handle it, and you don’t enjoy doing it—build the Custom GPT.

---

Building Your First Custom GPT

At this point, if you haven’t already made your first Custom GPT, you might be asking yourself how to start.

Let me share with you a couple of articles I published in the past that could be useful for you to understand more, and hopefully get some cool ideas.

---

Link one:

From Ping Pong Chats to Strategic Workflows: Custom GPTs and Actions

Link two:

Build a Custom GPT to Analyze Your Substack Data

---

You can skip the two previous links easily and just follow the following simple instructions:

My Prep Process

1. Start a regular ChatGPT chat

2. Define the problem clearly

3. Work with ChatGPT to refine a solution

4. End with a final prompt

5. Test it until the output feels right

Yes, I use ChatGPT to help me build Custom GPTs.

And I just learned that’s an Ingenic system.

The Technical Steps

1. Open ChatGPT in your browser: https://chatgpt.com

2. Click Explore GPTs in the sidebar

3. Click Create (top right)

4. Stay in Configure mode

5. Add:

   • Name

   • Description

   • Instructions (your tested prompt)

6. Click Create

💡 Tip: Store your prompt versions in a text document so you can roll back if something breaks.

It is possible that using it will help you find new use cases and features you didn’t consider before. Go back to the chat you used to create the prompt and work on the improvements. Once satisfied, bring the new prompt and replace the one in your Custom GPT.

Optionally, Ruben Hassid shows how he makes Custom GPTs in the following video:

---

What Comes Next (And What to Watch Out For)

Once you start creating your first Custom GPTs, you will find more and more applications. But before you start, I’d like to warn you of some risks:

Here, there is one important one by the best Jenny around :)
Jenny Ouyang has a particular way of expressing practical information about the best ways to use AI; she speaks to experts and beginners with ease. This one will apply when you use other Custom GPTs as sources of inspiration:

1. You will likely try to invent processes that have the potential to improve your work. I would keep mindful of “invention mode” because you could use far too much time creating solutions for problems that don’t exist.

2. Try using a diagraming tool to map your most important processes.
   I made a Custom GPT to help you on that.

3. ChatGPT is great at text and language tasks. More than giving tasks related to calculations or general research, think about applications related to working with text: interpretation, reformatting, classification, and analysis.
   Here is an example.

---

Timo points out a very important issue with Custom GPTs. Remember, the use of GPT should make your work straightforward, so you should avoid working on systems that you will not use afterwards. Chat GPT has some interface problems, and if you can’t easily find what you need, you might forget to use the tools you made with effort and time.

Sooner or later, you’ll have the battalion of Custom GPTs you dreamed about, but ChatGPT’s layout doesn’t organize them well.

ChatGPT shows you 21 Custom GPTs on the sidebar. When you cross that number, finding the right one becomes a frustrating scroll-and-guess game.

You might not have 21 Custom GPTs yet, but organization matters now because each GPT holds multiple documents in its knowledge base—voice samples, example outputs, formatting templates, process workflows.

When you need to update those documents, refine instructions, or add new training examples, you dig through ChatGPT’s interface to find the right GPT and hunt for the specific file.

A proper organization system stores all those knowledge base files in one place. Here’s the solution you can use from your first GPT.

I asked Timo how he organizes his Custom GPTs, and this is what he said:

I keep all my Custom GPTs organized in a simple Notion database.

What I track:

✓ Name — What the GPT does (e.g., “Article Architect,” “Hook Writer,” “Thread Repurposer”)

✓ Category — Email creation, short-form stuff, or business side-mission

✓ Link — Direct link to the Custom GPT in ChatGPT

✓ Knowledge Base Link — Direct link to the Google doc with examples or voice document

This system gives me two major benefits:

1. Faster access: I don’t scroll through ChatGPT’s sidebar trying to remember which GPT does what. I open Notion, search for the task, click the link, and I’m in.

2. Better organization: I categorize my GPTs by type (content creation, repurposing, email writing, etc.) and track which ones need updates or improvements.Save time, make your life easier.

I’m sure you could use other tools for the same purpose:

• Trello is free and can present you with a visual list of your GPTs.

• SharePoint lists can also be helpful; they usually come as part of the Office 365 suite.

• Nerd Alert: You could use Pivot Tables on Excel or other spreadsheets to easily find links to your favorite custom GPTs.

---

Some Custom GPTs I Love

There are so many amazing Custom GPTs available, but I would like to share with you two that I frequently use:

• CV Coach, by Andrew Stetsenko . It will revise your CV and suggest changes based on real-life experience with recruiters. It is amazing.

• SmartBook Fear Enough, it is trained with the full content of my book about fearlessness, learning from failure, and embracing entrepreneurship.

I also use a few I made that are not public at this moment, but maybe they inspire you to build something similar:

• Substack Note Optimizer: It analyzes the notes I want to publish, identifies them using a set of note types, identifies the purpose, and rewrites them in alignment with a set of priorities.

• Substack Subscription Growth Analyzer: I upload the CSV files with my subscription data, and it gives me all kinds of useful details about what is going on with my subscribers. This one is available for paid subscribers.

---

Wrapping up

Timo (Me) is taking over here… :)

Big thanks to Jose Antonio Morales for this collaboration.

- Just imagine I say his name with a perfect Spanish accent here :) -

Follow his work by visiting his profile and publications.

As you saw in this article, Jose knows his stuff. :)

---

Liked this post? 2 ways you can support me

1. Share it with a creator friend who will benefit.

2. Ask a question or share your thoughts in the comments.

See ya soon

Timo Mason🤠