AI Agency Runs on Just 5 Layers: NestJS, Supabase, Next.js, n8n and One API Call
soul_eater0001 · reddit · 2026-09-02
A developer running an AI automation SaaS agency shares the 5-layer stack behind every client project for years. Key thesis: job postings list capabilities, not tools—learn the layers and every new tool reveals itself as a wrapper.
- Layers: backend + database + frontend + runtime + AI call; every hyped tool wraps one of these
- Backend: NestJS (TypeScript, full products) or FastAPI (Python, AI-heavy); queues, retries, idempotent jobs, full logging—same scalable architecture for a 12-person client and a large one
- n8n: demos only; production is custom code hitting clients' CRM/helpdesk APIs directly—80-node n8n flows become undebuggable
- Database: Supabase (Postgres + auth + dashboard); vectors stored in Postgres for RAG, no separate vector DB
- Frontend: Next.js + shadcn—component code lives in your repo so coding agents can edit it; claims pure frontend dev is over, internal dashboards now take an afternoon
- AI layer: just an API call—go direct to Anthropic or OpenAI; everything around it is decades-old software engineering
More from coding & agent
- Merge launches enterprise AI governance tool enforcing model routing and spend rules — shensi · 2026-09-02
- Weaviate Ask Mode Adds Configurable Evaluation to Trade Latency vs Verifiability — CShorten30 · 2026-09-02
- Building the Ultimate Agent Harness for Kimi K3: The Model Is No Longer the Bottleneck — VibeMarketer_ · 2026-09-02
- GLM 5.2 slug references spotted in Google Antigravity CLI, hinting at integration — gaganghotra_ · 2026-09-02
- Design lead ships 12 PRs in a week: AI is erasing the designer-engineer gap — talkaboutdesign · 2026-09-02
- Measuring MCP servers: 78% of fixed token cost comes from tool schemas — Lexeik · 2026-09-02