## Demo 1: Default Output (Architecture Blueprint) **Model:** Claude 5 Sonnet **Input:** - PRODUCT_IDEA: AI habit coach SaaS — daily SMS nudges, adjustable targets, streak freeze, dashboard - USER_SCALE: 0 to 5k users in 6 months, 500 DAU peak - TECH_PREFERENCE: Next.js + TypeScript + Supabase preferred, avoid DevOps - TEAM_SIZE: Solo founder + 1 freelance QA - LAUNCH_TIMELINE: 4 weeks to beta, nights + weekends **Output:** # SaaS Architecture Generator *Product: AI habit coach | Scale: 0-5k, 500 DAU | Stack: Next.js + Supabase | Team: Solo + QA | Timeline: 4 weeks beta* ## Executive Summary **Architecture:** MODULAR MONOLITH **Confidence:** HIGH Next.js monolith + Supabase (Postgres + Auth + RLS) + Vercel + Stripe + Twilio covers 5k users with zero DevOps. Buy Auth/billing/SMS, build only habit logic. Modular folders keep future split easy without microservice overhead. **Build vs Buy:** Buy Auth (Supabase), Billing (Stripe), SMS (Twilio). Build habits, nudges, dashboard only. --- ## 1. Stack Decision (For Next.js + Supabase + solo) ### 4-Layer Stack | Layer | Choice | Why | Alternative | |---|---|---|---| | Frontend | Next.js 14 App Router + Tailwind | Solo knows it, Vercel deploys | Remix | | Backend/API | Next.js Route Handlers + Supabase Edge | No separate server | FastAPI | | Data/Auth | Supabase Postgres + Auth + RLS | Zero DevOps, RLS for tenants | Clerk + Neon | | Infra/DevOps | Vercel + Supabase + GitHub Actions | Free tier to $25/mo | Fly.io | ## 2. System Diagram (Text, for habit coach) ``` [Web/PWA] -> [Vercel CDN] -> [Next.js App] -> [Route Handlers] -> [Supabase PG/RLS] -> [Stripe] (billing) -> [Twilio] (SMS cron) -> [Claude API] (coach) ``` ### Integrations | Service | Purpose | Build vs Buy | Cost | |---|---|---|---| | Supabase Auth | Login, RLS | Buy | $0-25 | | Stripe | $9/mo subs | Buy | 2.9% + 30c | | Twilio | Daily SMS | Buy | $0.0079/msg | | Claude API | Coach msgs | Buy | ~$30/mo | ## 3. Data Model (3-5 tables for habit coach) | Table | Key Fields | Notes | |---|---|---| | users | id, email, phone, timezone | Supabase Auth + profile | | habits | id, user_id, name, target, freq | RLS by user_id | | checkins | id, habit_id, date, status | Index (habit_id, date) | | freezes | id, user_id, count | Earned 1/7 days | | messages | id, user_id, body, sent_at | Coach log | ### Multi-tenancy - Single DB + user_id RLS (not tenant_id) — B2C single-user, fits 5k scale. No schemas needed. ## 4. Build vs Buy (For solo + 4 weeks) | Feature | Build | Buy | Verdict | |---|---|---|---| | Auth | 1 week | 1 hr Supabase | Buy | | Billing | 1 week | 2 hrs Stripe | Buy | | SMS | 3 days | 2 hrs Twilio | Buy | | Habits CRUD | 3 days | — | Build | | Coach AI | 2 days | Claude API | Build thin | ## 5. Non-Functional (For 0-5k, 500 DAU) ### Performance / Security / Cost - Cache: Next.js ISR for dashboard, no Redis yet - Security: RLS on all tables, service key server-only - Cost: Vercel $0 + Supabase $25 + Twilio ~$20 + Claude $30 = ~$75/mo - Backups: Supabase PITR on Pro ## 6. Phased Build Plan (Fits 4 weeks) ### Week 1-2 - ☐ Supabase schema + RLS + Auth + landing + habits CRUD ### Week 3-4 - ☐ Checkins + freeze + Twilio cron + Stripe + dashboard + QA pass ### Launch - ☐ Beta 50 users, analytics (Posthog free), on-call checklist ## 7. What to Defer to v2 - Redis, separate API service, multi-region, custom admin, team workspaces ## 8. First Action Today - Scaffold: `npx create-next-app@latest` + `supabase init` + create habits/checkins tables with RLS