How are people pricing their AI apps?
Youssef El Mahallawy
8 replies
Ever since the release of ChatGPT's API, there has been hundreds of AI apps being launched every day. A lot of these apps have a flat-fee pricing plan, but I'm curious to how that was chosen, because every user will have varying token usage and it's hard to find an average token usage before you even launch your app. Some apps have pay-as-you-go pricing, so a user will pay for the amount of tokens they consume, but I feel like that adds a lot of unnecessary complexity to the project.
I wonder what strategies developers are using to pick a fair price for their AI applications. Anyone have any ideas?
Replies
Johnberg Arslan@johnberg
AppManager by CompanyDNA AI
We decided to charge the users separately from their ChatGPT usage. They enter their own API keys from OpenAI and pay for us. I think it is also possible to charge usage-based (like credits etc.) I am not sure if there are any other pricing models.
Share
AppManager by CompanyDNA AI
@james_tedy That's a really good question, I've been working at Gusto where we built API integrations with 60+ apps. There we made sure that every step of the flow was very clear and easy to understand for users so that they could set up their integration pretty quickly and easily. You need a good tutorial/education center to guide the users at every step of setup. Also if you can do the demo, you can show by yourself as well!
hi @johnberg , I saw this pricing models a lot for services that use ChatGPT underneath it.
I have a question about it, how do you handle customers that are not tech savvy? so they want to use the product, but the concept of getting API keys from somewhere else is new to them. (They don't understand it).
Wouldn't it deter them from using your product?
MyExpiring.Link
@james_tedy @johnberg What about for applications that use GPT-4? The app would break if someone who doesn't have GPT-4 access inputs their API key, unless there's a different version of AI for both customers?
AppyHigh Prime
To give X free trials so the users know what they are getting, and are happy to pay after the exhaustion of limit.
A lot of it comes down to the value of the service they've created. I know this from my own product I've launched. The value is in what it does rather than what the usage cost might be. There are a lot of other factors that also dictate the price plans. Things like hosting and other resource costs, staff costs for admin and future development. What often happens is that some users get an amazing deal because they're paying a flat fee and actually using a lot of resources, while on the other end of the spectrum there are some that are using few resources (or in some cases none at all as they forgot they're still subscribed). So somewhere in the middle is the average. And of course there's also profit to think of. The business R&D needs to be recouped as well. Lots of factors. The good thing is that entirely automated services can generate passive income.
A lot of it comes down to the value of the service they've created. I know this from my own product I've launched. The value is in what it does rather than what the usage cost might be. There are a lot of other factors that also dictate the price plans. Things like hosting and other resource costs, staff costs for admin and future development. What often happens is that some users get an amazing deal because they're paying a flat fee and actually using a lot of resources, while on the other end of the spectrum there are some that are using few resources (or in some cases none at all as they forgot they're still subscribed). So somewhere in the middle is the average. And of course there's also profit to think of. The business R&D needs to be recouped as well. Lots of factors.
MyExpiring.Link
@blissinvestor But it's really easy to get the pricing of an app wrong. It could be too expensive (because you have to account for the user's AI usage while also making profits) where it actually turns away users because they don't want that much for your product. On the flip-side, you can make it too cheap and end up losing money when your users run up your AI bill. How do you mitigate those consequences?