What 12 makers on MakerMap learned about it, in their own words: their best posts on X, filed under Building & shipping.
My stack for shipping an app (imo):
- design: Figma for structure, then straight to code
- build: Claude Code in the terminal
- App Store submit: all CLI, no clicking around
- discipline: run every day since January 2022, and ILTY for mental wellness because I needed it myself
My whole stack for FolioPress + Pickadate:
- Python + FastAPI (on my own little framework)
- SQLite, one file per site
- Jinja2 + Tailwind + BasecoatUI
- Docker + Traefik, blue/green deploys
- Cloudflare in front
No Kubernetes, no Postgres cluster, no React. Two small servers and it's plenty.
If you want to start a startup:
Claude = coding. ($20/mo)
Supabase = backend. (Free)
Vercel = deploying. (Free)
Namecheap = domain. ($12/yr)
ProductBridge = customer support + feedback (Free)
Stripe = payments. (2.9%/transaction)
GitHub = version control. (Free)
Resend = emails. (Free)
Clerk = auth. (Free)
Cloudflare = DNS. (Free)
PostHog = analytics. (Free)
Sentry = error tracking. (Free)
Upstash = Redis. (Free)
Pinecone = vector DB. (Free)
Total monthly cost to run a startup: ~$20
Nasab's privacy rule in one line: you can only read a row if you belong to that family.
It's enforced in Postgres (Supabase RLS), not in the app. The rest of the stack:
📱 Flutter + Riverpod
⚡ 17 Edge Functions
🖼 Firebase Storage
💳 RevenueCat
three products, one stack, zero decisions:
> next.js 15
> supabase
> tailwind
> vercel or railway
i reuse the same skeleton every time. the boring stack is why i can run three of these alone.
Built Vexavibes on Lovable for six months. Had to leave. Not a takedown. A warning.
We were building a trust platform. The box we sat in had openings we could not see until we owned the database. Signup email and sign-in kept coming back under new causes. The expensive failures reported success. We also had to close live beta accounts we had promised to keep.
Small app? Go. Something people have to trust? Own the backend on day one.
Article:
Company write-up:
I’ve built projects in WordPress, Shopify, React, Next.js and custom stacks.
My rule is simple:
WordPress when content is the product.
Shopify when commerce is the product.
Next.js when the software itself is the product.
Not perfect, but it prevents a lot of stupid decisions.
If you're tired of hosting your apps/web-SAAS on vercel/supabase, try this:
Browser / Mobile App
↓
Cloudflare
↓
Hetzner VPS
↓
Docker
↓
Nginx / Caddy
↓
Backend API
↓
PostgreSQL
↓
Redis / Queue
↓
Object Storage
GitHub
↓
CI/CD
↓
Docker Registry
↓
Server
Then add:
Backups
Monitoring
Logging
Secrets management
DB migrations
Staging environment
Rate limiting
Authentication
Cron/background workers
Testing
Rollback strategy
DobroDesk's support chat inside a Capacitor app uses SwiftUI on iOS and Compose on Android. It's native UI over a slot in the web page.
Hide it, reopen it, and your draft and pending messages are still there.
Why do EdTech platforms lag? It’s usually a bottleneck in the "trust layer." 🏛️
At #EduCreds, we use NestJS microservices + Azure Bicep to decouple verification from issuance.
See here👇👇👇
How it works under the hood:
⚡ Next.js (App Router): Client + API routes in one clean full-stack repo.
🎬 Remotion v4: Video as a pure function of React components f(frame) -> React tree. Headless Chrome captures frames + FFmpeg stitches the MP4.
🛠️TypeScript for type safety
Part 2 of my series on pragmatic engineering choices for solo founders and small teams.
You probably don't need Redis, RabbitMQ, or Kafka at your scale. The database you already run can carry your background jobs in a single table.
HEY pushes 20 million jobs a day through one. Your few thousand won't even warm it up.
And the industry's fix for the broker's lost-job problem? The outbox pattern: a jobs table in your database. The cure for the second system is the first one.
Count your jobs. Divide. Ship the boring version. You're allowed.