Video summary

The Only Tech Stack You Need to Build Apps in 2026

Main summary

Key takeaways

Technology

Summary of the video’s tech-stack recommendations (for building apps in 2026)

The speaker (Andrew) argues that choosing the wrong tech stack can waste months of time. He proposes a “layer-by-layer” stack optimized around three criteria for 2026:

  1. AI-ready (works well with AI tooling)
  2. Edge-fast (fast loading, low latency)
  3. Type-safe (catch errors at compile time rather than runtime)

He then walks through each layer, explaining his “winner” choices and alternatives.

1) Language (winner: TypeScript)

Contenders / considerations

  • Python for AI apps (e.g., AI + Python)
  • Dart for mobile
  • Swift / Kotlin for native iOS/Android
  • Notes: Kotlin can also be used for iOS, not only Android

Why TypeScript

  • Because much of the modern web/AI ecosystem is JavaScript/TypeScript, he favors TypeScript for type safety.
  • He recommends sticking to one language when possible.
  • He dislikes back-end JavaScript, and prefers TypeScript-centric full-stack workflows for beginners.

2) Front end (winner: React 19 + Tailwind + Next.js)

Styling/framework contenders

  • Svelte (described as a “hipster choice”)
  • Vue
  • Angular (enterprise focus)

Winner choices explained

  • React 19 as the mainstream default for front-end development in 2026.
  • Tailwind included (even though he dislikes some Tailwind/CSS aspects, mentioning CSS specificity/naming as “hell”).
  • Next.js singled out as the AI-app standard.

Overall rationale

  • Popular and ecosystem-compatible, especially for AI-driven apps.

3) Mobile (winner: React Native + Expo, using TypeScript)

Mobile contenders

  • React Native + TypeScript + Expo
  • Kotlin Multiplatform (sharing code between iOS/Android)
  • He personally dislikes Expo, but still selects it.

Why React Native + Expo wins for him

  • More jobs for React Native than Flutter (he claims ~6x more), and also more than Kotlin Multiplatform.
  • Expo Go accelerates early development (“get started so quickly”).
  • He notes Expo can be painful in real production due to update/version issues.
  • Despite that, he still prefers the React Native + TypeScript ecosystem and the unified language advantage.

Key warning

  • He considers this the least “perfect” part of the stack and encourages viewers to share better experiences.

4) Back end (winner: Node + NestJS; Rust/Go for performance/scale)

Performance / scale note

  • If scaling/performance is the main concern and language switching is acceptable:
    • Consider Rust or Go
    • He personally prefers Rust

Winner for practical fit

  • Node (never “dies,” and AI-generated code often uses it)
  • NestJS for a structured approach that works well with TypeScript
  • Express as a simpler starting point, then move to NestJS as complexity grows.

5) Database (winner: Supabase, with Postgres for customization)

Contenders

  • Supabase
  • MongoDB
  • Postgres

Guidance

  • Supabase for quick app starts, especially because it includes authentication.
  • Mentions “MCP server is amazing” as a reason to stick with Supabase unless you need deeper customization.
  • If you want more customization/complexity: use Postgres.
  • MongoDB: dismissed as having little place in modern stacks (after using it in the past via MERN).

6) AI layer (model API + orchestration)

Model / API choice path

  • Started with OpenAI API, but wasn’t satisfied with the experience.
  • Switched to OpenRouter:
    • One API for multiple models
    • Fallbacks and automatic switching based on cost
    • He calls it the cheapest option he found

AI orchestration choices

  • LangChain + MCP for coordinating AI behavior inside the app.

7) “Supporting cast” tools (recommended defaults)

Auth

  • Supabase auth for basic needs
  • Clerk for advanced auth features
  • Strong rule: don’t build your own auth (he claims even senior engineers get it wrong)

Payments

  • Stripe (default)
  • Mentions RevenueCat for mobile in-app purchases

Hosting

  • Prefers Vercel (especially with Next.js)
  • Mentions Cloudflare, but still chooses Vercel

Code editor / tooling

  • Cursor as his favorite editor (mentions “agents windows” and using Claude inside Cursor)

Analytics

  • PostHog

Error tracking

  • Sentry

Testing

  • Jest

Email

  • Resend

CI/CD

  • GitHub Actions

8) Product/tooling alternative: Emergent (sponsored)

He highlights a sponsored tool called Emergent, positioned as an “all-in-one” workflow:

  • User describes an app in plain language
  • It builds, deploys, and hosts front end, back end, and database from one prompt
  • Produces a production URL immediately
  • Platform handles SSL, scaling, and uptime monitoring
  • User can export code to GitHub with one click (not framed as a locked black box)
  • Presented as a faster path to shipping than debating a “perfect” stack

Final “winner” tech stack he lists

  • Next.js + TypeScript + Tailwind + Postgres
  • React Native + Expo
  • Supabase
  • Stripe
  • Vercel
  • Plus: Prisma (mentioned as important)
  • Supporting tools: Cursor, PostHog, Sentry, Jest, Resend, GitHub Actions

He concludes this combination is:

  • Learnable enough to get jobs
  • Fastest to ship
  • Scales without rewriting everything

Main speaker/source

Andrew (the channel’s creator; introduced as building apps for ~6 years and sharing his recommended 2026 tech stack).

Original video