my first B2C app made $700 in total. not $70k. $700. i spent about six months on it, and it turned out to be the most expensive education i'll ever get. five lessons that cost me six months, so maybe they cost you less. 1. entrepreneurship is a game of leverage, and your time is the weakest kind. when you're young you think: i'll give 100% of my time and that's enough. it's not. that math doesn't work. you need stacked leverage. co-founders. money. a real network. being fit and presentable. communication skills, being good on camera. every point of leverage multiplies the others. i had one. my time. it bought me $700. 2. think bottom-up, not top-down. beginners build like they're already a giant. i burned months on questions like: what if i get 10k users on day one? what if the db leaks? i can't use supabase for this, it needs to be on private DNS in AWS. all of it was premature optimization. it stole my time from distribution and my mental energy from decisions that actually mattered. build a beautiful UI and an embarrassingly simple backend (cloudflare stack is plenty). get your first customer. then ten. then a hundred. hire when customers are knocking, not before. 90% of your energy goes to distribution. everything else is noise. 3. mental energy is the currency. and vibe coding makes it dangerously easy to go broke. the trap sounds smart. five ideas, five prompts to five agents, multiply your chances. except the prompt is 1% of the work. shipping means tens of thousands of micro-decisions per app before it's production ready. do that across five apps and you will burn like a forest fire. my rule now is simple. max three projects, one project per day, rotate the days. all the ideas, none of the burnout. 4. distribution is the product. most beginners get this backwards. they ask: what good product can i build, and then how do i distribute it? flip it. ask: what distribution can i crack, and what product can i sell through it? if you don't have leverage like living in SF or having money, distribution is the only leverage you can manufacture from zero. go to local ecommerce events and sell a shopify app there. start a youtube channel about a topic and sell the app to that audience. the distribution decision comes first. the product decision comes second. pick a channel you can actually own, then build what fits through it. 5. sell it before you build it. a landing page and a waitlist cost one weekend. building the full app cost me six months. if ten people won't leave their email for the idea, ten thousand won't download the app. validate the demand first, then build only what the waitlist is begging for. take home summary. 1. stack leverage. your time alone is the weakest kind. 2. build bottom-up. beautiful UI, simple backend, get one customer first. 3. guard your mental energy. max three projects, one a day, no context switching. 4. distribution first. pick a channel you can own, then build what sells through it. 5. sell before you build. emails before code. the app made $700. the lessons were worth far more than that. building again now with all five fixed.

Missing the times when the $20 Codex plan used to give us like infinite GPT 5.6 Sol High 😭

It looks so void of nature—the devil's ultimate goal is to separate humans from nature, to make them like 🐄.

Someone is running out of money

i ship react native and native ios/android, and store owners keep asking which one they need for their shopify store app.

wrong question. the answer is a flowchart.

if the app lives and dies on checkout speed, with the apple pay sheet up in one tap on bad mall wifi, go native. that's the one screen a store app can't afford to stutter, and it's exactly where react native taxes you.

if the store already runs on react and the app is mostly catalog plus cart, go react native. one team, one language, the math works.

if the owner just wants 'an app' and has no engineering team, the honest answer is neither. ship the pwa and put the money into the email list instead.

pick wrong and you're paying two teams to sell the same products.

ai generated apps are about to flood the app store. but there is one wall that doesn't get cheaper with tokens.

app review.

apple bounces cookie cutter apps daily under guideline 4.3. same template, new icon, generated screenshots, copy that reads like a committee wrote it. i've had builds rejected for stuff that felt arbitrary at the time.

an agent can ship a working app in a weekend. it can't survive the 4.2 minimum functionality argument, redesign your onboarding because a reviewer decided you're just a web view, or write the appeal that gets the rejection reversed.

so the flood is coming, and then the flood is bouncing. apple's reviewers don't care what your token bill was.

spent 6 months building my first app. $700 total. writing the code was the easy part. getting anyone to care was the actual job.

expo gets the demo. swift gets the product.

openai ran out of computers for its $200 plan. bullish or bearish for the ai gold rush?

2023 me: writing swift all day. 2024 me: reading ai-written swift all day. 2026 me: defining tasks, setting permissions, approving output. same job title, completely different job.

openai just told $200/mo customers they can't take their money right now. their capacity broke before demand did. compute is the scarcest resource in tech and this is the proof. wrote it up

is $500/mo for ai tooling just the new normal, and we're all pretending it's fine?

every shopify store has a mcdonald's ice cream machine. the manual task the owner does every week and hates. that's the app idea, not another discount popup.

is it just me or has claude gotten kinda annoying to talk to lately?

i know devs with 10 years of swift who still can't tell you when not to go native.

what's your daily driver right now: claude, astra, or gpt sol?

honest question: do your users actually care if your app is native or react native? i've shipped both and not once has a user mentioned the framework. pretty sure the whole debate is just devs fighting in the replies

genuine question: is anyone still hyped about gpt-6 astra or did the hype quietly die already?