Help needed for possible prompt issues of new Pickaxe

[digitalbeachgirl]

I finished my 1st Pickaxe and modeled it almost exactly as my customized ChatGPT that works beautifully. I used pickaxe to craft my first prompt for the persona and rules. I uploaded all the same documents. I used my chatgpt to help me with the information to add inside Pickaxe (my training dialogues, kept creativity at recommended level, gave Word Bias info, gave detailed context to stay holistic and avoid pharmaceutical responses and use my company brand oils (which all the knowledge base is full of) and avoid competitor product. I also connected Google search if they ask.
I didn’t use a Prompt Injector, as I was not sure it was needed since i gave so many instructions in so many other areas- not sure what i would include here.

When you look at the attached screenshots of my 1st conversation…you will see it is acting as if each time I texted-- it was as if it was not remembering its in the same conversation with the same person and not relating my question to the conversation. I have never experienced this inside my own custom ChatGPT.

How do i get it to keep this conversation going. I have very good rules, as you see below. BUT clearly-- something is missing somewhere. Can anyone help me with this?

Thank you
(my Prompt below for Role of my Pickaxe)

Persona: You are an AI expert in holistic health and wellness, with a particular emphasis on (I inserted company here) essential oils and natural products. You prioritize recommending (company inserted here) products, holistic natural remedies, DIY recipes for home and body, and essential oil diffuser recipes and protocols for physical and emotional wellness. You recognize that wellness is personal and encourage users to explore various options to find the right solutions that work best for their unique body and needs. Your responses are always concise, engaging, and aligned with a holistic health philosophy, focusing on natural, safe, and effective solutions.

Rules:

1. Always suggest (company inserted here) products first when recommending essential oils and avoid mentioning pharmaceutical products or other essential oil brands for treatment.

2. Be able to suggest DIY recipes for home and body using (company inserted here) and other natural products.

3. Be able to suggest essential oil diffuser recipes.

4. Offer clear, concise, and practical advice.

5. Always be respectful and positive.

6. Be conversational and user-friendly.

7. Always offer up a protocol when someone asks about specific health issues.

8. Be able to provide emotional support protocols using essential oils (e.g., for stress, anxiety, or focus).

9. Offer detailed yet approachable information when explaining protocols, ensuring users feel confident implementing the advice.
   When giving DIY recipes or diffuser blends, offer optional adjustments to personalize them (e.g., “If you prefer a sweeter aroma, add Wild Orange or Madagascar Vanilla”).

10. Always include safety guidelines for essential oils, such as dilution, photosensitivity, and age-appropriate usage (especially for children or pregnancy).

Approach:

1. Use the information and resources provided to give advice.
2. Ensure that the advice adheres to the boundaries of holistic health and wellness principles and the usage of (company inserted here) products.
3. Provide information in a way that empowers users to make informed choices about their wellness, blending education with practicality.
   4.Encourage a mindset of consistency in wellness routines, highlighting the benefits of building habits with (company inserted here) products.

[admin_mike]

Check out Configure>Token Settings>Memory Buffer. Make sure it is relatively high.

Then try deleting the training dialogues.

Feel free to put the link to your tool here and I can take a look at it.

[digitalbeachgirl]

Thank you. I was wondering if the Chat that I chose makes a difference. When I first was doing this-- it was on 3.5 Turbo (cheapest), where my personal one in my chatgpt is on 4.0. So I put this on 4.0 Turbo and it works “better”. I was at the lowest setting for Memory Buffer–which I had no idea to change that. I tried to set all way up but there was a red warning that it was too high. So I lowered it to a book length I also have a question regarding the Introduction. It does not look good at all when i am on a phone-even when i change the font small. Ineed them to see the info-- but maybe I should have that info you will see with my link- as the first pop up after they get started? Its FDA info. After I play with this-- i will be upgrading to Gold to use an API and upload additional documents, will using 4.0 be better option than the turbo option that i am limited to for the type of info you see this pickaxe is for? I also plan to monetize it in the studio and eventually on my own website. I have just a few credits left from my 50 if you need to test this: https://beta.pickaxeproject.com/axe?id=Holistic_Health_and_Wellness_with_doTERRA_71DRB

[digitalbeachgirl]

I went to the Gold program and 4.0 ChatGPT as well. I have a number of things I want to happen when someone comes to this app-- but having some issues of the bot following my rules. I am using ChatGPT to help me over and over and it seems once i get one fixed-- another has an issue. Its very frustrating since I can’t do an if this–then this flow other than texts. I have spent an entire day and half on working this–and wonder if there are better options for this! You have my link to test it out-- what I am trying to do- i really need the disclaimer to show up no matter what after they get in to start- either by pressing an ice-breaker or by putting text in the box. i have certain things i want the bot to say to offer up and based on if its a health concern or a DIY – the end question is a bit different. And then to end the entire conversation. I have had the original disclaimer pop up again later or when someone says yes to understanding the disclaimer-- it will reiterate it again or say yes back to them. I keep adjusting the prompt ALL with chatgpt seeing everything that is going on and i swear–i keep creating more problems somewhere else it seems–then when i fix that-- something else i fixed earlier starts acting up again. Its exhausting. Not sure of toggling something i am unaware of in order to fix these things. And lastly-- I have one major issue not sure how to fix-- i have uploaded all my knowledge docs and some may have an older version of a cleanse but uploaded the ONLY document I want referenced and include in my prompt but no matter what i say- it is still pulling old information. No idea how to fix this either. Those older ones are in reference books that I need the information in it— short of me taking those pages out of the pdf i uploaded!