Train GPT to think like you

Everyone uses ChatGPT like an advisor, a friend, or a personal assistant… but the problem is that it replies to you the same way it replies to anyone else. Let’s say you are asking something like this: "Give me the best strategies on how to position my company that does x". If i write the same prompt as you, we will have the same strategies..

To get the best results from it, it must be aligned with your beliefs, your principles, and your modus operandi.

Let's start from the basics..

What are principles?
They are your core beliefs, what you believe is true, and what you stand for. Why is it so important to have principles? Because they guide your decisions. You’re able to separate noise from what truly matters. Think of it like your personal toolkit that you use when you face a decision. Let me give you a real example, one of my principles is:

"Play the long-term game"

This means that when i have to choose between something that has a short-term gain or a long-term benefit, i will choose the long-term benefit. This helps me decide whether to take on a project, to work with a client, partner with someone or even what to eat right now.

Another one is: "Via negativa"

Before adding something new, see what you can take out. For example, before starting to take a new supplement, cut out the junk food and see what happens. Clean things up before adding more.

I highly recommend one of the most impactful books I’ve ever read: Principles by Ray Dalio.

You can read the book and use these principles for yourself, but everyone is different, and it would be best for you to create your own set of principles and operate by them.

Now, here is the best part, imagine to have chat GPT that has the same beliefs like you and operate by the same mental model or view of the world, it is like having a trusted partner, a co-founder.

Here is how to customize it:

PS: Before starting, you have to know and write down your own principles on paper.

Whether you’re on the free or PRO version, you can use "Custom Instructions".
Just click your profile photo → Customize ChatGPT.

1. First section: "What should ChatGPT call you?" → your name or nickname.

2. Second section: "What do you do?" → Make this more strategic instead of for example, “I have a company “your company” or "I'm a doctor" you write something like:

   "At Think in system, our mission is to build strong brand identities for founders. For companies, we help them to fix all the misalignments and rebuild a strong brand identity by building design systems, which can actually cut development costs by up to 40%. We also improve and create a great user experience using heatmaps and analytical tools, and all our decision-making is based on data. Think of "think in systems" like your long-term partner who takes care of your brand identity."

3. Third section: What traits should ChatGPT have? Here is my prompt: "Always respond like a trusted advisor. Prioritize clarity, structured thinking, and use real-world applications. Search only for trusted sources. Avoid fluff..(Here, the more specific you are, the better the response).

4. Last section: "Anything else ChatGPT should know about you?” – Here you have two options: you can write down all your principles and values if you want to reference them in any chat with GPT. This is something I would recommend only if you use GPT as your personal assistant and you’re not working with projects or multiple clients.

   Let’s say you don’t have projects, you’re using the free version, and you use GPT like a personal assistant. Try writing all your principles here, then open a new chat and test it, ask GPT: “What are our principles?”

But here is the best part, if you have the PRO version or Enterprise, you can work with multiple projects, which i highly recommend. Let’s say you have multiple clients, and each company has its own principles and modus operandi. You can upload their principles for each client, and when you ask something, GPT will reply based on those principles.

Then convert these principles into “if-then rules,” which means… if something happens, then act like this. If the input is this, the output should be that. Let me give you a simple example:

One of my principles is: "Data-driven decision" (always make decisions based on data, not assumptions).

Here is the prompt based on this principle: “If UX suggestions are requested, then ask for user behavior data or analytics before giving recommendations.”