You’re not being accused of anything here my friend 
I honestly don’t think OpenAI wants to have their training data from ChatGPT filled with PII either 
But the builder of the GPT is allowing information to be sent anywhere, the 3rd party is now giving the builder information. So the third party is involved, and should be informed. I thought this was already done with the whole privacy policy thing anyway?
Ha! It’s turning into a gold mine for draaaamaaaaaaaa 
After all the comments I am no closer to understanding my original question.
Any thoughts on how one might go about monetizing custom GPT’s built through openai, that might actually yield some significant revenue?
Do it like Khanmigo is doing it.
On the GPT store, they had Khanmigo lite. If people like it, the GPT says they should check out their premium chatbot.
This is the answer, in my opinion.
I would not rely on the OpenAI revenue sharing, mainly because there’s been no reliable information released about it, and it’s isolated to USA.
Data harvesting is a low-hanging fruit that needs to be done carefully, so
Comments like this really degrade the warnings that people on a developer forum are trying to portray and truly provides no value.
Typically people need to be given value before expecting value back. This is a universal rule. People have already given value to OpenAI by subscribing to ChatGPT plus. If your GPT has advertising and/or data harvesting then it will be beaten by similar GPTs that provide the same functionalities.
If your GPT has some sort of moat, mainly by using Actions, then the value is in a premium subscription.
I’m having a tough time accessing my assistant through my WordPress site. Do you know if it’s easier to accomplish by taking Wordpress out of the equation or am I just a dumb dumb and missing something obvious? I have no problem using the regular API and one of the models to chat with through Wordpress (without using a plugin) but not that damn assistant.
Love your thinking, I’m in the process of setting this up right now actually.
See: OpenAI API
Now we’re just reinventing the wheel with extra steps
Check out this plugin. I use it and it really works quite well. AI Engine: The AI Plugin for WordPress
I’ve built a few GPTs that people have come straight out and told me, they would pay for the service my GPTs offer! Straight up, and that’s without me saying anything.
So I’m looking to put a pay wall on some of my GPTs but in a way that allows the user to test it, use it and see if it adds value to their lives.
Now the Pay wall I’m currently setting up at point of transaction requires nothing from the user other than a 5 digit alphanumeric key and a passcode. I don’t need their email address to be sent, I don’t need their IP address, they location or their blood type, and definitely not their pronouns, etc. Just their 5 digit key and their passcode.
My system is based off a credit system, you buy credits on my site, you use those credits to get access to ChatGPT GPT services. Simple.
Monetization is far away, unless you pay for direct use of the API, you buy tokens you sell tokens.
At this point, I get the impression that gpts are only looking for training for the MOE architecture of chat gpt, at this point the business I perceive is that if you directly manage the api, you set up chat bots with sexual content and make money, but that is not part of the specialized training for gpts, this is where the gtps of the store come in, the goal is to specialize the gtps to make the model more efficient and not call in all the experts. So while we see where this is going, in the future monetization will be valid as long as a valid business model is implemented.
For now I perceive that the overall model is going to be degraded with an emphasis on gpts, a free model that only activates the natural language module and then for a fee you will purchase subscriptions for custom gpts. At this point, we just have to wait and see if OpenAI sees this as profitable for gpts develo
You should make a GPT that explains how all this works.
I’m finding that with the voice capabilities of the ChatGPT I can “call” it and have a conversation, do my research while I walk the dog 
It’s great.
…I’m just lazy to read the site right now TBH
Im working on this strategy now for 2 of my GPTs. A developer I am looking at told me it would be better to create a custom version of my GPT on a website and then charge for access. Does this make sense? Ideally Id like to simply charge for access so that my customer could have a “Chatbot” or “AI Assistant” on their site by connecting to my API. Are these the same things?? And how hard of a task is this as the quotes for this work are coming in at a very wide range.
Sorry, “non-developer” here, but how difficult is this to do? And how does that work with OpenAI? If you set up a pay wall, clients pay you but is there any cost to you for use of the OpenAI API?
Just tested it out, didn’t do to well.
These are the options:
-
You create a GPT in the GPT store, then setup a universal type of pay wall that makes it easy for people to pay to use your GPT and other GPTs. This is currently what I’m setting up right now as we speak (or as I type) so my friends and myself can accept credit type of payment for our GPT’s. Initial testing has shown it can work, and it looks sweet! Main point, I’m not paying for people to use my GPT so other than my $20 a month to OpenAI plus my time, that’s my only cost.
-
Create an assistant, upload my knowledge to it and access through an API. You have more control over this, but with more control comes the requirement of way more skill to setup and host it. @basil0312 this is what your asking about. You will probably get quoted on the $1000 to get this done and rightly so, this skill/expertise including prompt engineering is a highly sort after highly paid skill. Plus infrastructure has to be setup to host the Ai service plus it wil require other AI services like vector DBs (but saying that, OpenAI assistants have built in RAG so vector DB is not required but I’m sure your getting hit on the price for it). RAG means uploading docs, PDFs, etc to a vector DB so your bot can search it and get answers to use in its response to your questions and tasks. You then run a chatbot on your website that connects to your OpenAI Assistant through an API call and you then own the whole thing but you will pay fir every call the api makes and that will add up so you have to have a pay wall in so it’s worth your while.
-
This is number 2, but others pay a big whack to lease your service on their website which is something I’ve only recently started looking into.
Now option one costs nothing for me to run, and I have loads of GPT plus users already to market to.
Option 2, I have setup a no code, low code way of creating and running assistants online fully under my control using and referring to my private data sources and I can connect it to any website and if you have a WordPress website, I can setup a pay wall as well. I’m currently now teaching a few people to do this themselves but this way is not cheap and does cost a bit of money.
Me personally, I got both running, well option 2 running and in the process of setting up option 1. When I finish option 1, I think it wil be useful to a lot of people.
@basil0312 give me a PM if you want to talk.
By the way this is my GPT made I made for my wife.
You upload your CV/Resume and a job description, and it generates personalised questions based of a set I tasks I trained it to do on both the JD and the CV/Resume with review and other outputs as well.
This is the one I’ve been told I should pay wall which got me started on option 1.
… plus my GPT wil email directly from this GPT, all the answers and advice once it’s done. This is code I’ve built into my GPT.
No, first, someone has to be able to find your GPT to begin with. How are they going to do that? As far as I can tell it’s impossible, there is no discovery tools in the store, notice I didn’t say ‘good’ I said ‘no’ - so I think before you can make money, people need to find you…
The so called search option seems extremely weak! But as we have seen with OpenAI things can improve in a short space of time.
Most people are depending on outside sources and marketing to get their GPTs discovered inside the GPT store. You might have a GPT that is well known and got a 1000’s of hits but when you look at the code, it’s nothing special (if you know what your doing you can rip the code and PDF/word knowledge docs as well to see the code behind the GPT). This is how I know some of these well used GPTs lack imagination. But there are some that are decent as well.
I’ve got a TL;DR that is better than most the TLDRs with big hits, functionality wise, but mine has not got as many hits to show on the board.
Check it out and feel free to give feedback, but this is another one I’d like to pay wall after letting users test it out 1st.