Is it profitable to build wrapper apps?
A wrapper app is an app that "wraps" a (usually) popular service. Like ChatGPT.
AppFigures, an app analytics company, recently wrote an article titled: Are 3rd Party ChatGPT Apps Still Making Money? The goal was to identify whether apps that basically wrapped ChatGPT's functionality still made money.
Yes in the short run: According to the article, Ask AI and Genie were two of the most popular and within a very short period of time, were making a lot of money:
According to AppAnnie's estimates, Ask AI earned $3.8M of net revenue in April. Genie made $1.7M in April.
I don't know about you, but that's a lot of money.
Things decline steeply in the long run: Ask AI's $4.6M of net revenue in August dropped to $3.9M in September. Genie's 2.1M in May has now shrunk to just $400K. So as ChatGPT released their own mobile app, these app fell in popularity.
How they promoted those apps: Mainly paid ads. According to AppAnnie, A big component of the growth of both was paid ads across many different channels, and that seems to have subsided in recent times.
To conclude, creating a "wrapper/more user-friendly interface" for a popular platform is definitely something worth doing if you want to make quick $$$. That is, until the official platform builds one on its own.
I think the key element when launching your wrapper is to target a very specific niche and also make sure this niche wouldn't be able to reach the output quality of your product by using ChatGPT alone.
Some times, you would need to refine your prompts for hours and manually input your external data to chatgpt to extract a response, this is where your product can shine.
Make it easier / faster for a niche to use chatGPT and take some money along the way :)
Nearly everything is a wrapper until you get down to language features and even then it's turtles all the way down.
We're just carbon wrappers
underrated.
Bruh that's the only true response
Yes, they were making a lot of money but they were also spending a big chunk of than on the OpenAI API and advertising. How profitable were they?
Depends on whether the wrapper can build a moat around it or not? For example, if your wrapper also contains data that will basically improve the model more so than a generic model could. Obviously this depends on that kind of product and market you're serving.
What if you train that LLM with something niche specific. Would that be a good moat?
Definitely, but the Moat must be defensible, meaning that it should be difficult if not impossible for someone else to create that dataset and experience.
I think the secret is to niche down, and also be creative. What I mean by that is to try and make something can't be easily replicated using chatGPT alone, an example of this is using multiple AI agents that will work together to generate a specific output.
Amazing
Rather than ask, is it a wrapper or in-house tech, what are the API incentives at play? We saw this with Facebook turning off their API and killing businesses overnight a few years ago after the Cambridge Analytica scandal, and we saw that when Elon took over Twitter.
Even putting aside the question of OpenAI or other API providers building their own solution to compete with yours, the big question is, could they increase API call prices or turn off the service completely? With an infrastructure provider, the answer is, "no," that's their core business. AWS isn't going to stop its cloud service, for example.
I've seen an argument that wrappers are a great business model. After all, a shop is just a "wrapper" in front of the wholesaler, a wrapper on the distributor, and so on. But the key in that equation is that each rung of the ladder specializes in a niche and wants to de-risk its business from the other rungs. Wholesalers aren't interested in running shops. Is that the case with OpenAI or Twitter allowing API calls?
This reminds me of Apollo for Reddit.
Wrapper apps are essentially MVPs, giving startups a chance to test the waters before diving deep into building something in-house. As for building in-house, it's essentially IP you build based on learning from customers, not necessarily a full in-house solution. It's just about being lean on each stage.
It depend on how you use it. For me creativeblogtopic.com I use the chatgpt mainly to analyze search traffic data and analyze keyword usage on competitor article . Instead of creating my own nlp to do this job, I found using chatgpt is cheap and better result.
And using function call you can really trim down your data like a real nlp model or any statistics formula you can use to analyse search engine data.
The idea is using the chatgpt exactly like you will use any other algorithm
Not if that is you core product proposal. Eventually GPT API will provide all functionality and whatever you build will be obsolete.
The profitability of building wrapper apps depends on various factors, including the target audience, functionality, and business model. In some cases, wrapper apps can be profitable by providing convenience or unique features, while in others, they may face challenges competing with native apps. Success depends on careful market research and a well-defined strategy.