"Minimum Viable Product" has become an excuse to ship garbage. That's not what MVP means. An MVP isn't the smallest thing you can build. It's the smallest thing you can build that _teaches you something valuable_. If your MVP doesn't answer a critical question about your business, you're just building slowly. This playbook walks you through shipping a real MVP in two weeks—one that real users can use and that generates real learning. ## Prerequisites Before starting this sprint, you need: ✓ Validated problem through user interviews ✓ Tested prototype with at least 3 users ✓ One clear hypothesis you're testing ✓ Technical founder or developer on the team ✓ Two weeks of full-time availability If you're missing any of these, go back. This sprint is for building, not discovering. ## The Two-Week Structure ### Week 1: Core + Happy Path Build the absolute minimum that works when everything goes right. ### Week 2: Polish + Launch Prep Add enough polish to avoid user confusion and prepare to launch. ## Week 1: Core + Happy Path ### Day 1-2: Set Up & Data Model **Day 1 Morning: Tech Stack Setup** Use the fastest stack you know—but TypeScript is mandatory: | If You Know | Use This Stack | | ----------- | ------------------------------------------- | | React | Next.js 14 + TypeScript + Vercel + Supabase | | Node.js | Express + TypeScript + Railway + PostgreSQL | | Python | FastAPI + Railway + PostgreSQL | | No-code | Bubble or Webflow + Airtable | **Critical:** Use TypeScript for all code. AI tools work better with types, and you'll catch bugs faster. Don't learn a new framework during an MVP sprint. Speed beats perfection. **Day 1 Afternoon: Database Schema** Use AI to accelerate: I'm building [describe your MVP] for [target users]. The core user flow is: 1. [Step 1] 2. [Step 2] 3. [Step 3] Generate a database schema that includes: - All necessary tables - Relationships between tables - Key fields for each table - Appropriate data types - Indexes for common queries Use PostgreSQL. Keep it simple—this is an MVP. **Modern Approaches:** For Next.js 14+, consider: - **Server Actions** instead of API routes (simpler, typesafe) - **Prisma** for type-safe database queries - **Zod** for runtime validation Example with Server Actions: \`\`\`typescript // app/actions/tasks.ts "use server"; import { z } from "zod"; import { prisma } from "@/lib/db"; const taskSchema = z.object({ title: z.string().min(1), description: z.string().optional(), }); export async function createTask(data: FormData) { const validated = taskSchema.parse({ title: data.get("title"), description: data.get("description"), }); return await prisma.task.create({ data: validated }); } \`\`\` This is faster to build and more type-safe than traditional APIs. **Day 2: Auth & Database Setup** Don't build auth yourself: \`\`\`bash # For Next.js npm install @clerk/nextjs # For Node.js npm install @clerk/clerk-sdk-node # For Python pip install clerk-backend-api \`\`\` Set up your database with one migration: - Create all tables - Add foreign keys - Add critical indexes only **Critical:** End Day 2 with a working dev environment and database. Test authentication. ### Day 3-4: Core Feature Implementation **Focus rule:** Build only the features that test your hypothesis. **Example: Task management app** Hypothesis: "Freelancers will pay for a task manager that automatically tracks time." **Must build:** - Create tasks - Start/stop timer - See total time tracked - Weekly summary **Don't build yet:** - Projects/categories - Team features - Reports/exports - Integrations - Mobile apps **Day 3:** Core data creation (forms, basic validation) **Day 4:** Core data display (lists, detail views) **AI-Assisted Development Strategy:** 1. **Use Cursor or GitHub Copilot** for real-time code completion 2. **Let AI generate boilerplate** (forms, API routes, database queries) 3. **Review and refine** AI-generated code (don't accept blindly) 4. **Use AI for tests** after writing core logic Use AI for boilerplate: Create a RESTful API for [your entity, e.g., "tasks"] with these operations: **Create:** [fields needed] **Read:** List all + get by ID **Update:** [fields that can be updated] **Delete:** Soft delete with deleted_at timestamp Stack: [your backend framework] Include: Input validation, error handling, TypeScript types Database: [your database] **AI Code Review Prompt:** After generating code, use this prompt to review: Review this [component/API/function] for: 1. **Security issues:** SQL injection, XSS, auth bypasses 2. **Performance problems:** N+1 queries, missing indexes, inefficient loops 3. **Type safety:** Any `any` types that should be specific 4. **Error handling:** Missing try-catch, unhandled edge cases 5. **Best practices:** Does this follow [Next.js/React/your framework] conventions? \`\`\`typescript [PASTE YOUR CODE] \`\`\` Provide specific suggestions with code examples. ### Day 5: Critical User Flow End-to-End **Morning:** Connect frontend to backend **Afternoon:** Test the entire flow yourself Can a user: 1. Sign up / log in 2. Complete the core action 3. See the result If yes → You're on track If no → Fix it before Day 6 ### Day 6-7: Polish the Happy Path **Day 6 Morning: UI Components** Use a component library: \`\`\`bash # Install shadcn/ui (recommended) npx shadcn-ui@latest init npx shadcn-ui@latest add button card input \`\`\` Replace basic forms with proper components. **Day 6 Afternoon: Loading States** Add spinners or skeletons where things load: - Page loads - Form submissions - Data fetching **Day 7: Basic Error Handling** Add user-friendly errors for: - Form validation ("Email is required") - Failed API calls ("Something went wrong. Please try again.") - Network issues ("Check your internet connection") **End of Week 1 checkpoint:** You should have a working product that handles the happy path. It's ugly and limited, but functional. ## Week 2: Polish + Launch Prep ### Day 8-9: Essential Edge Cases **Day 8: Empty States** What does the user see when they have no data yet? Bad: Blank screen Good: "No tasks yet. Create your first task to get started." **Day 9: Success Confirmations** After every action, confirm it worked: - "Task created successfully" - "Profile updated" - "Email sent" Use toasts or notifications (shadcn/ui has these built-in). ### Day 10: Beta Testing **Morning:** Deploy to production **Afternoon:** Test with 3-5 real users Give them a specific task: \`\`\` Scenario: You need to track how much time you spend on client work this week. Task: 1. Create 3 tasks for different projects 2. Track time on each 3. View your weekly summary Please think aloud as you go and tell me anything confusing. \`\`\` **Critical:** Watch them use it. Don't help. Note every point of confusion. ### Day 11: Fix Critical Issues From your beta tests, fix: **Must fix:** - Anything that breaks the core flow - Confusing UI that stopped users - Missing information users expected **Can wait:** - Nice-to-have features - Visual polish - Edge cases nobody hit ### Day 12: Landing Page & Onboarding **Morning: Simple Landing Page** Use AI to build it fast: Create a landing page for [your product] that includes: **Hero section:** - Headline: [your value proposition] - Subheadline: [what it does] - CTA button: "Start free trial" **Problem section:** - [The pain point you solve] **Solution section:** - [How your product solves it] **CTA section:** - "Ready to get started?" Style: Modern, clean, using Tailwind CSS and shadcn/ui Include: Mobile responsive layout **Afternoon: 2-Minute Onboarding** New users should reach value in 2 minutes: 1. Sign up (30 seconds) 2. Brief explainer (30 seconds) 3. Complete one core action (60 seconds) Example tooltips: - "Create your first task here →" - "Click the timer to start tracking ⏱️" - "View your time summary here 📊" ### Day 13: Analytics & Monitoring **Morning: Add Analytics** Minimal tracking: \`\`\`typescript // Track key events analytics.track("user_signed_up", { method: "email" }); analytics.track("task_created"); analytics.track("timer_started"); analytics.track("completed_first_week"); \`\`\` Use PostHog or Mixpanel (both have free tiers). **Afternoon: Error Monitoring** Add Sentry: \`\`\`bash npm install @sentry/nextjs \`\`\` You need to know when things break in production. **Modern Observability Stack:** | Tool | Purpose | Free Tier | | -------------------- | -------------------------- | -------------------- | | **Sentry** | Error tracking | 5K errors/month | | **Vercel Analytics** | Performance monitoring | Included with Vercel | | **PostHog** | Session replay + analytics | 1M events/month | | **Highlight.io** | Full-stack monitoring | 500 sessions/month | Pick 2-3 based on your needs. Don't over-instrument early. ### Day 14: Launch Prep **Morning: Pre-launch Checklist** - Core flow works end-to-end - Auth and password reset work - Mobile layout doesn't break - Error monitoring is live - Analytics tracking key events - Landing page deployed - Test account created for demos - 3 users completed beta successfully **Afternoon: Launch** Post to: - Your network (Twitter, LinkedIn) - Relevant communities (Reddit, Discord, Slack groups) - Product Hunt (if appropriate) Focus message on the problem you solve, not features. ## What You Should Have After 2 Weeks ✓ Working product deployed to production ✓ 10-50 users (realistic for week 1) ✓ Data on how users actually behave ✓ Clear insights on what to build next ✗ A complete product (you're not done) ✗ Perfect code (that's fine) ✗ Every feature you imagined (good—build what users need) ## Real Example: Time Tracking MVP **Week 1:** - Day 1-2: Next.js + Supabase + Clerk auth - Day 3-4: Task creation + timer functionality - Day 5: End-to-end flow working - Day 6-7: Basic UI + error handling **Week 2:** - Day 8-9: Empty states + confirmations - Day 10: Tested with 5 freelancers - Day 11: Fixed timer pause bug + confusing time display - Day 12: Landing page + "create first task" onboarding - Day 13: Added PostHog + Sentry - Day 14: Launched to 30 people **Results:** - 18 signed up - 12 created at least one task - 7 used it for a full week - Average 23 minutes to first value **Key learning:** Users wanted _project grouping_ more than any other feature. Built in Week 3. ## AI-Assisted Debugging When you hit bugs during your sprint: I'm getting this error in my [Next.js/React/etc] app: \`\`\` [PASTE ERROR MESSAGE] \`\`\` Context: - What I was trying to do: [describe] - What happened instead: [describe] - Stack: [your tech stack] Relevant code: \`\`\`typescript [PASTE RELEVANT CODE] \`\`\` Help me: 1. Identify the root cause 2. Provide a fix with explanation 3. Suggest how to prevent similar issues **Pro tip:** Use Cursor's AI chat with your entire codebase context for faster debugging. ## Common Mistakes **Mistake 1: Building features nobody asked for** Every hour spent on features not related to your core hypothesis is waste. Your job is to validate or invalidate one specific thing. **Mistake 2: Perfect code** Your MVP code will be messy. That's fine. You might throw it all away after learning what actually matters. **Mistake 3: Waiting to launch** Launch on Day 14 even if it's rough. You need real users to learn anything meaningful. **Mistake 4: No hypothesis** If you can't complete this sentence, you're not ready: "This MVP will succeed if users [specific measurable behavior]." ## After Launch: Week 3+ **First 30 days:** Talk to every user Set up 15-minute calls with anyone who used your product more than once. Ask: - What were you trying to do? - What was confusing? - What's missing? - Would you pay for this? **Measure your hypothesis:** Did users behave the way you predicted? If yes → keep building. If no → figure out why before adding features. ## Your Action Items Validate you have completed discovery work (interviews + prototype) Write your one-sentence hypothesis for this MVP Block 2 consecutive weeks on your calendar (80% of time) Choose your tech stack (the one you know best) Set up accounts: hosting, auth, analytics, error monitoring Recruit 5 beta testers before you start building Create a public launch date (accountability) Prepare to throw away features that don't matter Two weeks. One hypothesis. Real users. Real learning. That's what an MVP sprint delivers. Everything else can wait.