SaaS Architecture Guide

From idea to production, in reading order: what to read at each step and why, with lessons from 80+ projects.

1. 1. Architecture decisions

   How to think about system architecture, choose the right stack, and design for scale from day one.

   • preact-components: a Preact component library without shadcn or Radix

     Why I own my UI components instead of using shadcn, Radix or MUI: a Preact component library in ten JSR packages, rendered on the server, keyboard handling written by hand, with a live guide.

   • Cost-Disciplined SaaS Infrastructure: Managed Cloud, Dedicated, or Hybrid?

     A founder-readable framework for choosing managed cloud, dedicated, or hybrid SaaS infrastructure based on workload, team, compliance, recovery, and total cost.

   • Code & Security Audit, With a Fix Roadmap (From $1,500)

     I check your app's code and security and tell you what to fix first, including apps built with Lovable, Supabase, Cursor or Claude Code.

2. 2. Building the MVP

   Going from napkin sketch to production. Real projects I've built.

   • SmartLite

     Built solo in 3 months; about 200 lamp poles in production since 2024.

   • FoodRazor

     Scaled across 10 countries and hundreds of restaurants; acquired in 2023

   • Corecircle

     Scaled to 200K+ users; acquired in 2024

   • Truth or Dare (DareChat)

     ~40K monthly active users — live at darechat.me with a Swagger-documented public REST API at darechat.me/api.

   • Roley — Make a Movie!

     Built the MVP end to end. The client changed course to an offline business just before launch.

   • Connectful

     Showed fast growth during the COVID-19 pandemic

   • All client work
   • Build: SaaS MVP, From Idea to Production (From $8,000)

     I build your product from idea to a live app on servers you own. Backends, AI features and MCP servers are built the same way.

3. 3. Infrastructure and cost

   Choose infrastructure from measured load, recovery needs, compliance, team capacity, and budget.

   • How I run production

     Repeatable deployment, observability, recovery, security, and cost control after launch.

   • rostok

     One-command scaffolder for a self-hosted homelab.

4. 4. AI integration

   Add LLM capabilities to your SaaS with production-grade architecture.

   • I tested Trail-code on my Claude Code agents. They ignored the map.

     Trail-code promises cheaper, faster coding agents through a code map.

   • Claude Code vs MiniMax M3: 30 days against 3 months, in merged pull requests

     I counted my GitHub output for three months on MiniMax M3 and my first 30 days on Claude Code: 19 times more merged pull requests a day.

   • Opus 5.5 vs Sonnet 5: the pricier model wrote my code for about half the cost

     I priced five days of my coding-agent transcripts across four Claude models: 203 PRs and 411 reviewer agents.

   • Building MCP Servers with Deno: A Practical Guide

     Why Deno is the right runtime for Model Context Protocol servers, the architecture I use, a working CalDAV example, and what I learned shipping four MCP servers in production.

   • How ChatGPT Can Help You Design System Architecture for Your Applications

     ChatGPT as your architecture copilot: generate system diagrams, compare databases, spot security gaps before they ship.

   • AI integration and MCP servers, built under Build: MVP (From $8,000)

     Custom MCP servers that let your AI assistant read, write and act inside your own tools — CRM, database, email, calendar, internal APIs — shipped as a self-hosted Docker image you own.