Ive recently heard about ChatGPT’s new feature called context connectors, where users can link directly with Google Drive or Microsoft OneDrive to interrogate documents.
Won’t this feature compete with the API’s file search tool? Why would I then develop my own using the API if ChatGPT already has it?
1 Like
You could ask why you should do anything at all…
Do it because you want to do it, do it because you think you can do it better, do it because OpenAI hasn’t released theirs yet and there is no telling if or when they ever will…
Or don’t do it.
I’m not your dad.
3 Likes
There is also no guarantee that what OpenAI builds will work in the way that you want it to.
It’s easy (especially in this space) to assume that anything you build will be obsolete immediately, which is easy to extrapolate into ‘why should I ever do anything anymore?’ I think this will only accelerate as the technology improves. If you don’t find any personal joy or enough of a personal use case to build this now, maybe you shouldn’t. It’s all about your priotities.
I would appreciate respect in this community. Your answer was inappropriate and does not add any value (elmstedt).
My question resolves around the issue of building products for clients using OpenAI Assistants technology. If ChatGPT offers the same product, then it makes it harder to offer a unique value proposition as the API is in direct competition with chatgpt.
You had two questions,
- Won’t this feature compete with the API’s file search tool?
No, they’re entirely different products for entirely different spaces.
- Why would I then develop my own using the API if ChatGPT already has it?
Because the assistants API is an entirely different product than ChatGPT. Presumably whatever you are building with the assistants API is more than a ChatGPT clone plus the file search tool.
But, I didn’t get the sense you were here for an actual answer to either of those questions and were more interested in just complaining about potential features being added to the consumer product.
2 Likes
If you are building something using the Assistant API, I would encourage you to continue your efforts. Even if OpenAI integrates similar features into ChatGPT, I don’t believe your efforts will be wasted.
Of course, it’s your valuable time, so prioritizing is necessary, but the experience gained from developing new features with the Assistant API will undoubtedly be beneficial in the future.
2 Likes
This is the aftermarket developers’ dilemma. They are often conflicted about investing in what is ostensibly a wrapper. @che.kulhan you should be conflicted. Building anything that matches what is central to a vendor’s product is risky and unlikely to establish clear differentiation.
To be clear, ChatGPT is a “demo” that stuck. It lacks security, privacy, performance, and many things customers cherish and will pay for. Your job is to identify your moat and the unique value you bring to the marketplace and build something clever that others would find difficult to match.
5 Likes
Thanks for your responses.
My overall concern is that although the API has awesome features like secure/private file search, it seems that ChatGPT may incorporate this type of feature in the future with the Connectors. Hence making its own file search tool obsolete.
1 Like
Yay! I heard about this and can’t wait for the feature. I’m excited. It solves a big hurdle for the ChatGPT UI, namely making API connections to my Google Workspace, which I needed yesterday.
Meanwhile, I am also learning to create APIs via Assistants, and to also create GPT Actions via CustomGPTs.
ChatGPT and the Assistants UI are for completely different audiences, for all that they both duplicate services on a 1-to-1 basis. ChatGPT is verbose and friendly, Assistants are direct and powerful. ChatGPT is the verbal solution, appropriate to large audiences; Assistants are the Developer solution appropriate to this audience.
You learn both just like @anon22939549 says for funizes or you don’t learn either like @pds says because everything you do will be obsolete immediately.
I’m learning both because I use both in my work, and I think both are valuable at different stages for use in different types of projects. They can’t render each other obsolete because they are the same product provided to different audiences. It’s a brilliant strategy.