I’ve developed a service called Gapier, which includes 26 well-adapted Action APIs that can be integrated into GPTs within 60 seconds, providing 26 additional capabilities. The current capabilities include:
Drawing-related APIs, such as various posters, statistical charts, and mind maps.
Information search APIs, including books, videos, audio, and movies.
Tools, such as email sending, search, and simple memory functions (clipboard).
I had this idea at the second week of the developer day and immediately began executing it, and it went live last week. However, I am unsure if this service has user value, and whether OpenAI will appreciate this service in the future. Developers, please provide some advice for the perplexed developer.
Do you also have something for the API devs or your APIs are specialized for use in GPT only? This is for those who want to make their own “custom GPT” using Assistants API.
hey, at the moment, i’m only focusing on cold-starting for GPTs users.
We are working hard to make it a universal capability for easier integration with all LLMs, Assistants, and Agents.
Impossible to say if or when the feature will be phased out. Being new, it’s likely to be heavily promoted initially, similar to YouTube Shorts or Instagram Reels at launch.
A common sentiment on this forum is truly differentiated GPTs come from Actions & API calls, which require a bit of coding knowledge. However, impactful GPTs are still achievable with just custom prompts and creativity, no coding needed.
Yes, I also recently came to a deep understanding of the concept that “truly differentiated GPTs come from Actions & API calls,” and the realization that non-developers may find it challenging to utilize Actions & API capabilities.
I have recommended gapier to some content creators and am seeking advice on whether I should target them as the main users. @Benito What would you do if it were you?
If you asked me on DevDay I would’ve said yes 100%. It seemed like OpenAI really is filling the engines for GPTs. Now, after the fiasco they seem to be sitting ducks. I still can’t believe there hasn’t been any sort of leadership or official announcement to say “We got this, stay tuned!”.
They really lost the momentum after the GPT store was delayed.
Eh. OpenAI has no attitude. Historically they hype things up and drop the ball a couple weeks later. They are as taken back as a company can be.
I think with great ideas such as yours non-developers will be given an equal playing field.
In my experience with OpenAI they tend to steamroll anybody/anything that tries to improve their products. Amazon Basics style. They gather statistics, release it themselves, come out strong, appeal to the masses, and then disappear, effectively killing the improvements.
The plugin marketplace was awful. Part of me believes they never intended for it to actually function, they just wanted to gather statistics of what people wanted so they could create it themselves. So I rightfully have my doubts about this GPT marketplace.
The lack of communication and historical trends of OpenAI to me makes any sort of improvement or technology that relies on their ecosystem dangerous.
Different models from different companies are gaining traction and catching up. You may want to consider building this marketplace in such a way that it can be used by them as well, and not completely rely on Actions.
Probably yes, as the intelligence and context limits expand and cost goes down, having a million different specialized functions will probably be quite useful.
On another note, I really appreciate this approach to advertisement.
i am also developing a product through gpt now.
Nowadays there are many company take part in the competition of LLM.GPT still keep a lead location.
Some plug-in is abondoned just because gpt-4-turbo’s strong ability. But there are also some plug-in still is useful,such as pdf-ai. So whether your product will be out or not depend your product’s irreplaceability
@pzdzxlx
I agree.
I believe Gapier is actually doing two things: enhancing some fundamental capabilities of LLM and connecting.
I’m still researching to figure out which is more reliable and useful.
I suggest you keep your plans short term as I suspect OpenAI is just using the api to gather ideas and will discontinue it when their product is more mature.
@blueeon I agree 100% with you, lets dm or connect over email to discuss more. You can find my contact at contact section of thumbnailai.online. I got your website name from reddit
On a most basic economic level there is generally an incentive for such things to be absorbed. A bit like how game mods can become part of the core game experience. How much OpenAI will respect us as modders essentially will become apparent I guess. I suppose the counter would be that they want modders to mod so have to generally look after the existing ones to some degree.
The concept of “GPTs,” which is quite a way to pervert the name by the way, and the entire concept of the assistant API, is such a knee-jerk reaction by OpenAI. It felt like something from a Y Combinator meeting to discuss how to create the next iPhone app store moment, but it’s actually falling flat on its face because, let’s face it, the expertise at OpenAI is not necessarily in creating value in this way. Shocker: What they are actually best at is creating the best large language model and other transformer-based models in the world.
I couldn’t agree more.
So far, anything at the application level seems inadequate, even ChatGPT itself.
Without a powerful model, who would be obsessed with a product filled with various low-level bugs?
But what cannot be ignored is that they are the leaders of this wave, and I think they may continue to evolve. After all, chaos is the norm.
I believe developers do not trust OpenAI enough, possibly because they themselves are unsure if what they are launching is useful. They just want to lanuch it to see the user’s feedback. After all, innovation cannot be planned.
I think the key is the form of GPTs, whether it can make LLM’s ability practical, and make users feel truly useful. OpenAI explores ahead, and developers follow at low cost.
Just my personal thought, and I’m not certain if it’s accurate. It’s open for discussion.
It makes a lot of sense to expect that the GPTs will change in the (near) future.
But luckily we can already draw from the experience with the Plug-in store:
From the developer’s perspective there were changes but it is definitely possible to port a plug-in to a GPT. The development efforts are not lost.
From a marketing perspective users were directed towards a plug-in and now this needs to change to direct them towards the GPT. Which is not the greatest but also not a major roadblock either.
So I do expect that efforts to develop GPTs can pay off. But one needs to be adaptive.
