The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 13 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Jiang: Spending three hours vibe-coding a prototype compounds significantly over time
“Sometimes you can just take a first step to getting started, whether it is three hours of vibe coding a prototype just to test something with users or writing your first like, Hey, I'm building this app and this is day one. That's all you really have to do som…”
Jiang: Self-improvement apps must guide user actions, not just track metrics
“It's the same kind of concept of, like, you know, you have to, like, capture not just the tracking and the numbers part. It's like there's so much more, so many more steps to ultimately getting to you taking an action and then seeing the results of the action.”
Jiang: Apply the 'Rule of Six' for optimal app prototype user testing
“So I had this thing I call a rule of six, which means talk to six people that you think is in your ideal customer profile. Two less is like, you don't have enough sample size, too many. It's like, you just don't move fast enough because like, you know, people …”
Jiang: Peek combines Claude for reasoning with Gemini for fast tool calling
“As for the LLM side, we actually string together a combination of, ah, Gemini, as well as Claude, ah, depending on like the, I guess the function, but a lot of the initial insights, like, These kind of I would say like more detailed insights. We use three seve…”
Jiang: Founders building in public must rely on replies before standalone posts
“When you're, you know, early on, like, reply a lot, and then at some point, your post will carry you more”
Jiang: Human taste remains essential because 90% of AI-generated ideas are terrible
“I mean, the one thing you still have to have is you have to have taste. I mean, they'll give you a bunch of ideas, like nine out of 10 are shit. You have to decide, like, which one, like, you know, catches people's attention, is visually suitable, and can rela…”
Jiang: Peek's initial prototype was coded using Cursor and v0
“I basically used cursor and V zero to code this out.”
Jiang: Peek uses a vector database to store thousands of user signals
“We have a vector database that actually stores memory from the user. So we have like actually thousands of signals that we save down, right?”
Jiang: Peek's cross-platform stack relies on TypeScript, React Native, and Expo
“So for like the front end and the back end, it's been in TypeScript. And then for the mobile app because we want it to be like more platform agnostic across, you know, iOS as well as Android use React Native and using Expo for that.”
Jiang: Peek users engage 7x more frequently than traditional desktop dashboard users
“We're already seeing how people are using our app nearly seven times more frequently than like your desktop dashboard app.”
Jiang: Peek doubled week-three retention using behavioral and onboarding reminder notifications
“We did two things that actually shot our week three retention up from that, ah, from like, 13% to essentially, like, 28% in, like, a short period. We improved onboarding completion, literally just asking people, hey, it's been three days, just connect your ban…”