My cofounder & I fight this a lot…
hear me out
We’re testing the market through web before investing heavily in native apps.
A study of 162,000+ iOS devices found that 90% of users regularly open only around 14–18 apps per week.
i think for a 2-person cofounder team, convincing someone to open a website feels much easier than earning a permanent spot on their home screen.
but Jerry disagree.
Curious how other consumer AI teams think about this.
Hot take:
AI inference is the new Silk Road.
hear me out
2,000 years ago, the Silk Road created price differences based on geography, access, and politics.
Today, AI models doing the same.
Kimi, Qwen, GLM… depending on where you source them, the exact same model can cost ~30% less compare to OpenRouter
WHAT is the long-term moat here: the model, or the distribution??
120k users in 3 days
Going viral too early feels more complicated than people make it sound.
hear me out
You get attention, but you also lose novelty and attract competitors before you’ve really figured things out.
@garrytan what’s your thoughts on virality?
anyone else experienced this??
Hot take:
Great products create great datasets.
If nobody wants to use your product, training your own model probably won’t solve that.
Distribution first. Proprietary models later.
@garrytan curious if you agree
@unfairdotso tbh I don’t think it was distribution. We accidentally hit a format people wanted to create, not just consume so the content became the acquisition channel.
Curious if you’ve seen similar loops elsewhere? @unfairdotso
My cofounder and I have been debating this a lot.
hear me out
we’re starting with web instead of building a full app first.
It just feels like the fastest way to test if people actually want what we’re building.
Curious how other consumer AI teams think about this.
@JimsonYang_ bruh if resources were unlimited I’d probably agree. But for a 2 person founding team (n we tend to keep the team as lean as we can), every extra platform slows product decisions too.
you ever seen a consumer startup successfully keep both moving at the same speed early on?
@liquidated0x 100%. I think iteration speed compounds way more than UI polish in the beginning. At what point do you think native actually becomes worth the cost?
My cofounder and I were talking about infra today.
We’re leaning toward keeping things simple early on and just calling APIs.
If users love the product, we’ll optimize later. If they don’t, better infrastructure probably wasn’t going to save us anyway.
Would love to hear different opinions.
If you're a founder, indie hacker, or building anything at all, let's connect. Reply with your project and follow. Building a network of people who actually ship.
WTF IS WRONG WITH SF?
I’m from Taiwan 🇹🇼 , 15 hrs flight away
my first time hosting a event in SF
and we got nearly 100 application
30 are selected to join this dinner
LOOK AT THESE ATTENDEES OMG
Please make sure you supermodels show up tonight :)