If I don't know with local knowledge, I just ask the hotel reception for any hospital/doctor where they speak English.
I think my health insurance provider can also provide me with contact info of relevant parties.
We ran into the same question with WIP where initially it was all based on Telegram usernames, but we then made having a Telegram account optional. Same with BetaList where we only used Twitter OAuth, but more recently added regular signups.
There isn’t a super clean solution.
What I ended up doing is letting the user choose their own username, but pre-fill it with their OAuth service name if available.
It hasn’t lead to any issues.
One extra benefit is that it gives people an incentive to sign up early to claim their desired username.
For what’s it’s worth I’m never a fan of this approach, as it temporarily exposes part of my email address to other users. Assuming the username is public.
How does it expose the email address? These names come from the creators' names on platforms like Twitch/Twitter/Discord. So twitch.tv/jaku my link would be mywebsiteservice.com/twitch/j….
I suppose if jaku is part of my email this is true, but they'd have to assume the rest of the email and what provider I'm using to receive email?
Unless I'm missing something here?
Being able to access GPT-4 anywhere seems super useful. However, for me those triggers words never feel that intuitive. I'd much prefer having auto-complete like Copilot provides. You just type and see suggestions you can tab to auto-complete. That feels a lot more natural.
As for the "roast", there's A LOT of competition in this market. And all apps seem to do roughly the same thing. I think it will be very hard to build a competitive business, if your main selling point is providing a keyword trigger for the OpenAI APIs.
I'd really niche it down to something specific (sorry, not sure what that should be). Or come up with a more streamlines UX that's harder to replicate.
The direction going forward really depends on your unique strengths and interests. I'd focus on those.
Having said all that, major props for shipping this! Most people get stuck at the idea phase. You actually delivered. Now all you need to do is keep going 🙌
+1 to what Marc said – the internet is flooded with products right now that all do some version of “We pre-prompt your prompt and give you no-code access to GPT API”
That’s good, iff you have a niche non-technical market you can reach better/cheaper/faster/moretargeted than competitors. Or as Marc said, you make the integration suuuuper buttery smooth for your target users in an existing app they’re already using.
For example: I would love a ChatGPT interface inside VSCode that understands the entire context of at least the local file, full project would be even better, and is as easy to use as TypeScript-based autocomplete.
But even if that existed, my head of security would never let me use it 💩
Thanks Swizec. I actually have a same idea, but not sure if I could make it better than Github copilot haha.
As of security/privacy, I think a paid opensource could work. WDYT? :D
Thanks a lot for your reply, @marc. And I agree with everything you said.
I gave myself 1 week to ship this product and would then collect feedback from my early customers. Hopefully I could find some niche use case that make this really useful.
Otherwise, yeah, it's quite hard to build a competitive business with the current feature.
There's a bunch of members using Stripe Connect. If you don't get an answer here, I recommend reaching out to them.
wip.co/search?type=&extra_que…
Thanks @marc will do so 🫶