Feedback - Discoverability of GPTs Needs Improvement

GPT builders Plugins / Actions builders
, , ,
1

Hi OpenAI team,

First off, congrats on the launch of the GPTs store. I think you guys did a great job simplifying what was originally intended in the mockups which is awesome.

My one piece of feedback for the store is I think the discoverability of GPTs is lacking. What I mean by this specifically is currently, only the top 12 GPTs get shown but there are tons of great GPTs that most of us probably have no idea about unless they were directly shared with us. On the current implementation, they are essentially completely hidden which when monetization comes around, essentially is pointless for 99% of the store.

As a suggestion, I would offer one of two things:

  1. Simply show us all the GPTs on a different page per category. So for example, say I click on the writing category, I think it would be a simple improvement to have a “see all” button that shows all the GPTs in this category, instead of just 12. From there, you could filter through “most used”, “new” etc.

  2. Add recommendations for users. This is of course much harder technically, but would make for a great experience for users. Say something like “Since you used this GPT, you might like this”. This down the road would be awesome.

I love the GPTs store concept and I’d hate for it to get stuck over something like this which seems like a bit of a shortcoming currently. Happy to provide more feedback if wanted.

Thanks again for the great work!

9 Likes
2

Get one of those top twelve and extract their instructions. Most do very little and are next to useless. I have no idea why they were chosen.

3 Likes
3

Agree some are basically just a prompt then off to their site…not to impressive.

7

I think we need a new “Just for fun” category

8

This is a great suggestion. I created a category just for fun, enjoy it!

9

I think that GPTs with APIs would be a great filter for users as they are inherently more powerful/useful beyond stock ChatGPT. That being said I’ve seen some amazing GPTs without APIs but I still believe it to be a very useful filter to surface. I’d use it!

:white_check_mark:API

10

I agree with both suggestions. But I also think that GPTs with added value (actions and knowledge files) should be promoted more than those that do not add anything new and are just popular because of their name.

11

Thank you to both for sharing your insights!

I’d like to share the latest categorization based on the keywords for API action, Web Browsing, DALL·E Image Generation, analyzing data with a code interpreter, and File-based Knowledge Retrieval. Besides adding new categories, I’ve also conducted the following statistics and assigned specific names to each category:

Total GPTs identified with these keywords: 12,660, with 6,759 categorized labels

  • API Action (named GPT API): 4.5%
  • Web Browsing (named GPT Browsing): 92.9%
  • DALL·E Image Generation (named GPT DALL-E): 85.5%
  • Analyze data with code interpreter (named GPT Code): 44.4%
  • File-based Knowledge Retrieval (named GPT Knowledge): 21.6%

Having all the above features: 0.5%

Additionally, for those categorized under File-based Knowledge Retrieval (GPT Knowledge), the corresponding number of files will be displayed for each GPT.

It’s important to note that the total number of GPTs is based on the current keywords and does not represent the entirety but can serve as a reference.

Enjoy it!

12

I agree entirely - just having 12 GTP’s in each category is worse than useless and no incentive at all for anyone to work on, or publish a GTP. How these top 12 are selected also needs improvement, as the selection is clearly not based on the quality of the GPT. Something pretty drastic needs to be done to improve the situation, if the GTP store is going to have any value at all. Imagine if the Apple store launched with a restriction of just 12 apps per category. I doubt it would be around today.