ZeroToShip

Screenshots

Automated SaaS that scrapes the web for technical pain points and generates prioritized business briefs for aspiring founders. Pipeline runs daily, delivers ideas by email, and serves a tier-gated web dashboard.

Live: zerotoship.dev

---

Architecture

┌───────────────────────────────────────────────────────────────┐
│  DAILY PIPELINE (cron, 5 AM PT)                               │
├───────────────────────────────────────────────────────────────┤
│                                                               │
│  Scrape (Parallel)         Analyze (Parallel)      Deliver   │
│  ┌─────────────────┐       ┌───────────────┐      ┌────────┐│
│  │ Reddit Scraper   │──┐    │ Deduplicator  │──┐   │ Email  ││
│  ├─────────────────┤  │    ├───────────────┤  │   │Delivery││
│  │ HN Scraper       │──┼──▶│ Scorer        │──┼──▶├────────┤│
│  ├─────────────────┤  │    ├───────────────┤  │   │  Web   ││
│  │ GitHub Scraper   │──┘    │ Gap Analyzer  │──┤   │Dashboard││
│  └─────────────────┘       ├───────────────┤  │   ├────────┤│
│                             │Brief Generator│──┘   │  API   ││
│                             └───────────────┘      └────────┘│
└───────────────────────────────────────────────────────────────┘

Tech Stack

Layer	Technology
Backend	TypeScript, Fastify, Drizzle ORM, Zod
Frontend	Next.js 15 (App Router), React 19, Tailwind CSS
Database	PostgreSQL (Supabase)
AI	Anthropic Claude (scoring + briefs), OpenAI (embeddings)
Email	Resend
Payments	Stripe (checkout, webhooks, billing portal)
Hosting	Railway (API + scheduler), Vercel (web)
Cache	Redis (ioredis) with in-memory fallback
CI/CD	GitHub Actions → Vercel auto-deploy
Analytics	PostHog
Monitoring	Sentry, pipeline watchdog, alerting (Slack/SMS/PagerDuty)

Monorepo Structure

src/
  scrapers/       Data collection (Reddit, HN, GitHub)
  analysis/       Clustering, scoring, gap analysis, embeddings
  generation/     AI brief generation (legacy + graph providers)
  delivery/       Email sending, onboarding drip
  api/            Fastify REST API, middleware, services
  api/db/         Drizzle schema and database client
  scheduler/      Pipeline orchestration, cron, watchdog
  config/         Environment, models, Redis
  lib/            Logger, Anthropic client, semaphore, errors
packages/shared/  Types shared between backend and frontend
web/              Next.js frontend (separate workspace)
tests/            Mirrors src/ structure
drizzle/          SQL migrations
scripts/          DB seed, Stripe setup, utilities

Quick Start

# Install dependencies
npm install

# Copy environment template
cp .env.example .env
# Fill in: ANTHROPIC_API_KEY, SUPABASE_URL, SUPABASE_SERVICE_ROLE_KEY, etc.

# Run database migrations
npm run db:push

# Start API server
npm run dev:api

# Start web frontend (separate terminal)
npm run web:dev

# Run pipeline once
npm run scheduler:run

# Start cron scheduler
npm run scheduler:start

Key Commands

# Backend
npm run build              # Compile TypeScript
npm test                   # Run backend tests (Vitest)
npm run dev:api            # Start Fastify API server

# Frontend
npm run web:dev            # Next.js dev server
npm run web:build          # Production build
npm run web:test           # Frontend unit tests
npm run web:test:e2e       # Playwright E2E tests

# Scheduler
npm run scheduler:run      # Run pipeline once
npm run scheduler:dry-run  # Dry run (no emails)
npm run scheduler:start    # Start cron scheduler
npm run scheduler:health   # Check environment
npm run scheduler:watchdog # Monitor pipeline health

# Database
npm run db:generate        # Generate Drizzle migrations
npm run db:push            # Push schema to DB
npm run db:studio          # Open Drizzle Studio
npm run db:seed            # Seed with sample data

# Validation
npm run validate-costs     # Validate AI cost estimates
npx tsc --noEmit           # Type check

Pipeline

The scheduler runs four phases daily:

1. Scrape — Collects posts from Reddit (28 subreddits), Hacker News (Algolia API), and GitHub (issues from 500+ star repos). 50+ pain point signal patterns.
2. Analyze — Deduplicates via OpenAI embeddings (cosine similarity 0.85), scores with AI + heuristic hybrid, runs gap analysis (SerpAPI/Brave + AI competitor research).
3. Generate — Produces structured business briefs via Claude. Two modes:
   • Legacy: Single Claude call per brief (default)
   • Graph: Multi-node pipeline with model cascade, section-aware retries, synthesis, budget controls. Enable with GENERATION_MODE=graph.
4. Deliver — Sends tier-based email digests via Resend and persists briefs to the database for the web dashboard.

Generation Modes

Mode	Model	Description
Legacy	Sonnet 4.6	Single-pass brief generation. Stable default.
Graph	Cascade (configurable)	Multi-attempt with critic evaluation, section retries, synthesis node, budget caps, trace visualization.

Graph mode is controlled by environment variables:

• GENERATION_MODE=graph — Enable graph provider
• GRAPH_MAX_ATTEMPTS — Max retry attempts per brief (default: 2)
• GRAPH_MAX_SECTION_RETRIES — Per-section retry cap (default: 1)
• GRAPH_MAX_CONCURRENT_BRIEFS — Concurrent brief generation (default: 5)
• GRAPH_RUN_BUDGET_USD / GRAPH_RUN_BUDGET_TOKENS — Run-level budget caps
• BRIEF_GENERATION_TIMEOUT_MS — API call timeout (default: 180000)

Tiers

Tier	Price	Features
Free	$0	3 ideas/day, problem + solution summary
Builder	$19/mo	10 ideas/day, full briefs, archive, priority delivery
Enterprise	$99/mo	Unlimited, API access, custom filters, export, history

Configuration

See .env.example for all environment variables. Key categories:

• AI: ANTHROPIC_API_KEY, OPENAI_API_KEY
• Database: DATABASE_URL or SUPABASE_DB_URL
• Auth: SUPABASE_URL, SUPABASE_SERVICE_ROLE_KEY, Google OAuth
• Email: RESEND_API_KEY, RESEND_FROM_EMAIL
• Payments: STRIPE_SECRET_KEY, STRIPE_WEBHOOK_SECRET, price IDs
• Scheduler: SCHEDULER_CRON, SCHEDULER_TIMEZONE, GENERATION_MODE
• Monitoring: SENTRY_DSN, ALERT_EMAIL, ALERT_SLACK_WEBHOOK

Testing

npm test                   # 1395+ backend tests across 55 files
npm run web:test           # Frontend unit tests
npm run web:test:e2e       # Playwright E2E (journeys, accessibility, edge cases)

Tests mock all external APIs. No real API calls in tests.

Documentation

Document	Purpose
Context.md	Current project state and working memory
Feature-Plan.md	Architecture, roadmap, competitive analysis
Strategy-Analysis.md	Market sizing, competitive landscape, 90-day plan
docs/architecture/	C4 architecture diagrams (context, container, component, deployment)
docs/planning/	LangGraph migration roadmap, decision log, KPIs
reviews/	Code reviews, plan review (16/16 complete)
reviews/career/	Interview-ready project evidence

Career Demo Docs

For interview-ready project evidence:

1. reviews/career/README.md
2. reviews/career/impact-scoreboard.md
3. reviews/career/experiment-log.md
4. reviews/career/decision-log.md
5. reviews/career/weekly-build-log.md
6. demos/README.md

License

MIT