Who has had enough of GPT-wrapping products?
Gabriele Mazzola 🧭
16 replies
While I can understand the excitement of building things with the use of GPT, I feel that this trend might "pollute" (pass me the term) the community and will make it hard to spot the products where the team actually puts a lot of effort into building.
Of course, some "wrappers" of GPT are still incredibly valuable, but I'm seeing way too many "products" just being an API call with a pre-defined prompt.
Replies
Alex Nix@n1x_ax
While there are hundreds of simple products on the market with a simple GPT wrapper, you can find products that are really worthy of attention. But since GPT-4, for example, has only recently appeared in general use, all these products are in the early stages of MVP or still in development, which is normal. I am sure that soon we will see many different worthy products, as people are still learning to use it in conjunction with other services and build rare automation.
The team and I also worked for 6 months on the GPT product, and we are only launching at the end of this month. But we have done a lot of work not to be just a GTP wrapper, but to create a really unusual development that improves the basic technology.
What are your thoughts about it?
Share
@nixkulinax I totally agree with you!
My team and I are also working on a product that - among several other things - uses GPT in the backend. However, in our case - and I guess in yours too - integrating and using GPT is only a fraction of the engineering effort that gone into the product.
My post was absolutely not meant to judge these projects, which I actually admire and enjoy, but rather those that are just basically making a couple of API calls to GPT and then return the response as if they suddenly became the masters of generative AI.
@gabriele_mazzola agree! Totally understand you!! What is the product you are building?
@nixkulinax I'll share more details the upcoming days, after scheduling the launch. But it's a system leveraging AI, human-curated content, and spatial data to provide a new immersive experience for travellers!
If you're interested in the topic, make sure to check on my page tomorrow as I'm planning to publish the teaser and the landing page for it :)
Honestly I've been trying to wrap my head around training custom models instead of wrapping my product in GPT. But I Need more knowledge on machine learning.
GPT gives a fantastic 60% solution, which I think is better than nothing.
It's like the Dotcom boom but with machine learning
@david_deisadze
Train the new LLM - $20-80M
Fine-tune existing LLM - $200-400k
Build the wrapper - $500 :)
@david_deisadze @thisisanton Still, I believe there can be dumb wrapper and incredibly smart wrapper! And I have nothing against the latter.
Launching soon!
😅. 
)
)I'm sick of numerous "AI content writers" or "AI chatbots" that are pretty much always the same.
But some wrappers are pretty good when it comes to a certain specific function.
And a good UI/UX and dashboard features.
Have patience... :) It is like when the iPhone started to support native apps and App Store. For a while the top grossing app was an app making farting sounds...
Now everyone is going after the low hanging fruit, being first to market. A majority of all those products will not get enough traction and will be gone in 12 months, and only those that bring actual value will be left.
I share your sentiment 🥲
Nowadays, many "products" are essentially just serving as a "shell" for pre-configured prompts rather than offering a fully-conceptualized product
I believe we should exercise restraint in using GPT. AI should fundamentally serve as an enabler that complements human efforts and actualizes our visions, rather than acting as an expedient pathway to an ostensibly polished product or usurping the role of human cognition and imagination