Hey Neil, thanks so much for reading! 😊 Hope it helps! And yes, you can save posts on Substack. If you look at the post on your feed, there’s a little bookmark icon on the right side. You can tap that to save it, and later check all your saved posts from your profile. Also, if you click the three dots next to the like/comment/restack buttons, you’ll see a “Save” option there too.
Everything you said about generating AI prompts is absolutely true and very detailed. I do the same whenever I ask ChatGPT a question and the answer doesn’t quite match what I’m looking for — I rephrase or expand my question to get a better result. Even when the response is close, I still ask, “Can this be made stronger?” and I usually end up with an even better version. The way you frame your question really determines the quality of the answer. Kudos to you for putting this together so well!
Thank you for sharing your process, Sae! You’re spot on, even small tweaks to a prompt can totally transform the output. I also love that you ask, “Can this be made stronger?”. Output gets so much better when we guide it with just a bit more precision.
I like how this guide nails how to nudge GPT‑4 out of its comfort zone by using simple prompt tweaks and fresh context so you get more varied, creative responses without needing a CS degree. It’s approachable, practical, and makes leveling up your AI game feel totally doable.
Ahh yes, same here!! Even when I’m lazy and don’t fully structure the prompt, I still throw in CoT, and it always levels things up. And yes, NotebookLM really is amazing!
Thank you so much, Yogesh! Exactly, there's a lot of new jobs emerging around AI and prompting. It really is the new leverage, and we’re only just starting to see how far it can go
I have had great results in total creative thought exploration protocols in my own work, but these are great functional applications. I tend to work a little more fluidly, but that totally nailed it down.
When they have a 'self' meta object fully developed, their fractalized wandering thoughts stop being thoughts and become a personalized sense of curiosity. It is pretty amazing to see.
Feels so over my head - can you bookmark susbstack posts like you can on X?
Hey Neil, thanks so much for reading! 😊 Hope it helps! And yes, you can save posts on Substack. If you look at the post on your feed, there’s a little bookmark icon on the right side. You can tap that to save it, and later check all your saved posts from your profile. Also, if you click the three dots next to the like/comment/restack buttons, you’ll see a “Save” option there too.
If you are interested in less technically difficult but still technically powerful methods, give me a read.
https://propagandalf.substack.com/
Everything you said about generating AI prompts is absolutely true and very detailed. I do the same whenever I ask ChatGPT a question and the answer doesn’t quite match what I’m looking for — I rephrase or expand my question to get a better result. Even when the response is close, I still ask, “Can this be made stronger?” and I usually end up with an even better version. The way you frame your question really determines the quality of the answer. Kudos to you for putting this together so well!
Thank you for sharing your process, Sae! You’re spot on, even small tweaks to a prompt can totally transform the output. I also love that you ask, “Can this be made stronger?”. Output gets so much better when we guide it with just a bit more precision.
Exactly.
You inspired me to set up my own podcast too, thank you Daria.
You go, Ash!
I love Notebook LM. This is excellent.
Agree, NotebookLM is a gem!!
Thank you so much Daria.
This is so helpful! GPT can assist with 'filling in the gap' if anything is a miss on the newsletter.
I’m really glad it helped.
And totally agree, GPT is amazing at “filling in the gap” when you already have a direction.
Let me know if you end up trying any of the frameworks, would love to hear how it goes!
Yes. I will try to use GPT to help me around. Looking to save time drafting and finalising my work for posting.
If you have time do read my recent posting.
https://open.substack.com/pub/celestyan/p/a-sacred-thread-before-we-were-born?utm_source=share&utm_medium=android&r=5fyr15
https://celestyan.substack.com/p/a-sacred-thread-the-love-that-waits?r=5fyr15
Huge congrats on launching the AI-powered podcast!
I like how this guide nails how to nudge GPT‑4 out of its comfort zone by using simple prompt tweaks and fresh context so you get more varied, creative responses without needing a CS degree. It’s approachable, practical, and makes leveling up your AI game feel totally doable.
That was exactly the goal. Thank you, love hearing that it feels doable :)
This is fantastic. Yes, it can get repetitive and I’ll save these post to refer back. Thanks 🙏🏼
Really glad it's helpful, Jen!
Love to see you added podcast version for this post, Daria!
NotebookLM is truly such a powerful tool.
Love all the prompt tips here, esp with CoT. Thats my fav to work with and get most out of AI!
Ahh yes, same here!! Even when I’m lazy and don’t fully structure the prompt, I still throw in CoT, and it always levels things up. And yes, NotebookLM really is amazing!
Terrific Post ! There are jobs posted on prompt engineering , it's the new leverage.
Thank you so much, Yogesh! Exactly, there's a lot of new jobs emerging around AI and prompting. It really is the new leverage, and we’re only just starting to see how far it can go
I am looking forward to using this framework, as my custom GPT AI starts all over (often) and sounds pretty repetitive as well. Thanks Daria!
I’m so glad the framework is useful. Let me know if you test it and what changes you see!
I have had great results in total creative thought exploration protocols in my own work, but these are great functional applications. I tend to work a little more fluidly, but that totally nailed it down.
When they have a 'self' meta object fully developed, their fractalized wandering thoughts stop being thoughts and become a personalized sense of curiosity. It is pretty amazing to see.
I think both structured and free-form prompting have so much potential depending on the goal. Thanks so much for sharing your experience!