What apps really cost, the fixed-price market with verified prices, and honest comparisons of the platforms and services around it. Written by the two engineers who do the builds.
Agencies quote $40,000 to $300,000 over three to twelve months. What actually changed in software economics, what did not, and what a launchable core really costs now.
Never the screens. The hard parts are decisions: what happens when two users act at once, where the money ledger lives, how a feed stays fast. Where difficulty actually lives, category by category.
Traction is the moment the shortcuts come due, in a predictable order: queries, then everything synchronous, then everything you cannot see. The sequence of fixes that works, and when to rebuild instead.
A toolbox, not a sacred stack: Next.js, Supabase, Clerk, Pusher, Expo, Stripe, the AI model layer, Render for long-running work, and the bench behind them, each matched to its job.
Eight fixed-price builders with their real published prices, from $1,999 to $60,000, verified on their own sites. Timelines, code ownership, support, and the fine print.
Yes, three ways: AI platforms, freelancers, and fixed-price senior teams. What each one produces, when each wins, and the arithmetic behind the agency red-flag warnings.
Where the platform ceiling sits, who actually builds a business on these tools, and why a real app on your own infrastructure is a different class of asset.
Eleven services that fix stalled AI-built apps, from €99 bug fixes to $150,000 takeovers, and the arithmetic on when a one-week rebuild beats a rescue.
The real engineering difficulties behind 15 well-known products, what a one-week core includes, and what we would leave out.
Or see how the week works.