
Goodbye Gen AI Wrapper Startups
OpenAI just released GPT 4 Turbo and and GPT builder, where you can just create a custom GPT in 5 mins. Insane!!
RIP to them cool tech bros who thought creating a wrapper startup is going to make them m/billionaires. Building a startup ain't that easy :)
Talking product sense with Ridhi
9 min AI interview5 questions

The competition cannot catch up to the pace at which Open AI is operating

How is this different from finetuning a model using the finetuning api's provided by them

It's different because now you don't have to use any APIs. Just interface

So they made a ui wrapper over their own api's to kill wrapper startups xD
Also you can use a ui interface to finetune a model in the openai playground right now aswell, so idk what this ui interface does differently

ELI5 version please

Letβs say there is an imaginary app which does the following:
You upload a hinge profile snapshot, it analyses everything in that uploaded snapshot and tells you what to respond to either a prompt or a pic so that you have the highest turnaround.
This app, under the hood(I.e backend of course) used OpenAI model, where you had given a list of instructions to generate a response considering various factors, for example, whether the person drinks socially/regularly etc, their job, their age, their gender and also your profile parameters.
Now with the latest release, you yourself can do the above mentioned thing using ChatGPT interface.

This doesn't usually affect solopreneurs since the task there is to make chatgpt accessible to folks who are not using/learning gpt.
Eg- even if there is dall-e, those interior design startups wont be affected since the ui abstraction is easy for interior designers to use.
Another analogy would be - why do people use paid software when there is open source version of almost everything - EASE OF USE.
Creating a Gen Ai wrapper is usually a team of 1-2 people. If you can generate 1k-5k Mrr from that, you can easily sit back since that passive income is more than enough
Another advantage of building is - you dont build only for the current product, you build to stay on top of latest changes. Eg for chatbot companies, it was easy to adopt gpt etc since they have everything else built already. Hence a chatbot building indie hacker might not have made much earlier but they instantly integrated gpt and are not doing thousand of dollars of MRR.
Building is important always and is almost always a step up in your career.
Sitting back and saying your gen ai wrapper is useless now is just counter intuitive and kinda petty also.

Now*

It's not that simple.
Tabnine is not dead even after two major release of GH copilot.
There is a vscode fork called cursor which plans to support LLM on user hardware.
Jasper and many gen AI text tools are still in business
Pricing : New features will cost at least 20$/month compared to many of those chat with pdf apps which cost way less. You can actually use bing to chat with your pdf for free. Even from API standpoint the new assistants API which provides retrieval functionality is although simpler to implement but still costlier than creating your own vector index.
Privacy : many orgs wont upload their data to openAI, they will rather self host a vector store. There are YC W23 startups working on private GPT for enterprises, even though openAI released their enterprise plan long ago.
UI : there are(where) multiple reddit clients with millions of downloads - all they do is use reddit API to provide better UX.
At the end of day companies & users don't care about how it's implemented under the hood. If it has better UX and integrations with their workflow they will continue using existing solutions, for that reason I believe chatbase founder will still earn millions of $$.
Personally I think this update is big only for casual users and small orgs. For everyone else dedicated tools are better.

OpenAI / Microsoft has deep enterprise relationships that will easily overcome the privacy and security concerns of Enterprises. See Slack vs Teams. Pricing: yes, but the question is build vs rent. Renting is cheaper, especially because of the pace at which the underlying product is improving - no tiny startup can hope to outpace that.

Oh and Reddit has effectively shut down the 3rd party clients already with the pricing changes on APIs. The few that will remain will not get the most support in the future, itβs only a matter of time before they are folded into the official apps.

Any obvious names?

This is just like Appstore or Playstore but only for AI based applications. So you can use it make your own application and sell through it as well.