← Back to SE Resources
Advanced Topics

🚀 Advanced Topics

The AI Cloud Layer, composable architecture, customer architecture patterns, and migration playbooks — the advanced knowledge that differentiates top SEs.

The AI Cloud Layer

Vercel's biggest strategic bet. As an SE in 2026, you need to understand and demo these tools.

Fluid Compute + AI = Cost Efficiency

LLM calls = mostly I/O wait. Traditional serverless bills for all wait time. Fluid Compute bills only ~10ms of actual CPU per call → 80-90% cost reduction for AI streaming routes.

Composable Architecture & Headless

Vercel positions itself as the presentation layer of a composable stack. A huge part of enterprise sales.

Frontend (Next.js on Vercel) connects to:

Headless CMSContentful, Sanity, Storyblok, Hygraph
E-commerceShopify, Commercetools, BigCommerce
SearchAlgolia, Elastic
AuthAuth0, Clerk, NextAuth
PaymentsStripe, Braintree
AnalyticsSegment, Mixpanel
MediaCloudinary, Mux

Common Customer Architectures

The patterns you'll encounter most in enterprise sales. Know how to draw and explain each.

E-commerce Pattern

Backend: Shopify / Commercetools
RouteStrategyDescription
/products/[slug]ISR (1hr + webhook on update)Product pages with on-demand revalidation
/collections/[slug]ISR (15min)Collection listings, periodic refresh
/cartCSR (client-side)Sensitive, user-specific cart data
/checkoutSSR (always fresh)Auth required, real-time inventory
Edge MiddlewareA/B testing50/50 homepage split, cookie-based

Media / Publishing Pattern

Backend: Contentful / Sanity
RouteStrategyDescription
/ (homepage)ISR (5min or webhook)Breaking news, editorial updates
/articles/[slug]ISR (webhook + tags)Tag-based revalidation on publish
/liveSSR (no cache)Real-time ticker, always fresh
/searchSSR or CSRUser-specific search results
CMS Webhook→ revalidateTag('articles')On-demand invalidation

SaaS Dashboard Pattern

Backend: Clerk/Auth0 + Vercel Postgres/Supabase
RouteStrategyDescription
/ (marketing)SSGStatic marketing pages
/blogISRBlog with CMS integration
/loginSSGJust a static form
/app/*SSR / RSCAuthenticated, user-specific data

AI Application Pattern

Backend: AI Gateway → OpenAI/Anthropic + Pinecone/Qdrant
RouteStrategyDescription
/chatRSC + useChat (AI SDK)Streaming chat interface
/api/chatVercel Function (streaming)LLM response via Fluid Compute
/api/searchEdge FunctionFast vector search at the edge
/api/agentuse-workflow (durable)Long-running agent tasks

Migration Patterns

You will be asked about migrations constantly. Know these playbooks.

Create React App → Next.js App Router

Common
  1. Add Next.js alongside CRA (incremental adoption)
  2. Move routes one by one
  3. Identify SSG vs ISR vs SSR per route
  4. Replace API calls with Server Components where possible
  5. Add Middleware for auth (replace route guards)
  6. Migrate to next/image, next/font, next/link

Gatsby → Next.js

Medium
  1. GraphQL data layer → Server Components with direct data fetching
  2. Gatsby static generation → Next.js SSG/ISR
  3. Gatsby plugins → Next.js equivalents or direct integrations

Self-hosted Next.js → Vercel

Easy
  1. Usually zero code changes — point DNS to Vercel
  2. Gain: ISR distributed globally (not single-region)
  3. Gain: Preview deployments for every PR
  4. Gain: Zero-config builds + Skew Protection
  5. Gain: Automatic image optimisation at CDN level

Pages Router → App Router

Incremental
  1. Migrate one route at a time (pages/ and app/ coexist)
  2. getServerSideProps → async Server Components
  3. getStaticProps → SSG Server Components
  4. Data fetching moves from page-level to component-level