New guidance posted for what makes a great ChatGPT app: a practical guide for capability choice and conversation- & discovery-first design
15 Likes
Awesome post! Very helpful!
this is awesome.
@corey when returning info to the UI widgets, do you have any advice/guidance on specifically where to return the “first meaningful response that explains the role in one line?
I’ve played around a little with the response text along with the returned UI+Content but the model doesn’t necessarily output exactly what’s returned.
Well,
technically speaking:
The post already lacks basic audit requirements:
Where is the design of the UI ↔ model return logic?
Where is the contract schema (e.g. JSON-RPC, Protobuf)?
Where are the differentiated layers (widget logic, content stream, fallback override)?
Where is the detection of invalidated output (e.g. BLAKE3 vs SHA256 mismatch)?
Where is the trace route renderer —> stream —> token decoder?
Seems like everyone want to build a great ChatGPT app.
Two questions:
- What does OpenAI get out of it with respect to monetization? After all, they need to raise 1.4 triilion to finance future data center and cloud infrastructure commitments.
- What do developers get out of it with respect to monetization? Will there be enough customers willing to pay for god knows how many ChatGPTs running around? This assumes, of course, that the great ChatGPT apps deliver as expected.
Developers really need to think about this - will the fruits of their labor and headaches associated with ever-changing models ever be worth it?
I can say that I truly don’t know. But I’m beginning to have doubts.
1 Like
Thank you for sharing @coreyching!
Can we build an App on other language too, or is it just on English possible?
I think, when we make together stronger the ethically and human safe side with GPT App’s, it’s can be very positive on the global market too.
What do you think about it?
One thing is missing from the ecosystem. How do you plan to let developers make money?
2 Likes