Choosing Your Stack

What This Is

Eleven platforms in. Every one of them is good at something. Every one of them has a homepage that will make you sign up in twenty seconds. That's the trap.

This chapter isn't an opinion. It's a framework. Four questions. Concrete answers. The tool falls out the other end. Whether you're a non-technical founder, a dev shopping for an upgrade, or a weekend builder with a credit card — once you've answered the questions, the decision stops feeling like a coin flip.

One thing up front: no stack choice is permanent. The best builders I know treat tool selection as provisional — right for now, revisited later. The goal isn't the perfect answer forever. It's the right answer for this project, this month, with a clear path out if you outgrow it.

---

Who Should Use This

Non-technical founders who've been paralyzed. You've watched the YouTube comparisons, read the Reddit threads, tried two platforms, shipped nothing. Ten minutes of this chapter and you have a place to start.

Developers who don't want to chase shiny objects. You know what you're building. The framework tells you where the leverage is — and where it isn't.

Teams making a purchase. You need a structured argument, not a gut feeling. The persona section gives you the vocabulary.

---

The Four Questions

Write the answers down. Everything else follows.

1. What's your technical level?

Not years of experience — three things. Can you read an error message? Do you know what frontend vs. backend means, roughly? Are you willing to spend time in a terminal?

• Three nos → Lovable, Bolt, or Vibecode. Browser. No install.
• Some yeses, willing to learn → Cursor, Claude Code, or Replit. Higher ceiling, more engagement.
• Experienced dev → Cursor (local IDE), Claude Code (terminal-first), or Copilot (enterprise). Pick by workflow taste, not capability.

2. How long does this need to last?

Determines how much platform risk you can absorb.

• Days to weeks → Lock-in doesn't matter. Use whatever ships fastest.
• Months — MVP that might become real → Think about export now. Bolt with download, Cursor, Vibecode.
• Years — production → Code you own. Infrastructure you control. Cursor or Claude Code.

3. What are you building?

4. What's your monthly budget?

Real spend right now — not annual ambition.

• $0 → Every platform here has a free tier. Start there. Hit limits. Reassess.
• $20–50 → The sweet spot for individuals. Pick one primary tool and go deep. Spreading $50 across three subscriptions is worse than spending $20 on one you actually understand.
• $100–200+ → Team or revenue. Primary professional tool plus the infra it needs. Cursor Teams + Supabase Pro + Vercel Pro is the standard coherent stack.

---

The Flowchart

START HERE
│
├─ Non-technical?
│   ├─ YES → Need a working app in days?
│   │           ├─ YES → [Lovable]
│   │           └─ NO  → [Vibecode] or [Bolt]
│   │
│   └─ NO → Already in an IDE?
│               ├─ YES → [Cursor] (+ optional [Copilot])
│               │           └─ Terminal-only? → [Claude Code]
│               └─ NO →
│                   ├─ UI components?  → [v0] → [Cursor / Replit]
│                   ├─ Full-stack?     → [Replit] or [Bolt]
│                   └─ Voice-first?    → [Wispr Flow] over any of the above
│
├─ Prototype? → free tier, ship fast, upgrade if needed
├─ MVP w/ growth potential? → prioritize export → Bolt, Cursor, Vibecode
└─ Long-term production? → [Cursor] or [Claude Code] → own your stack

If two paths point somewhere plausible, you're in the overlap zone. Try both on free tiers. The decision is preference, not strategy.

---

Stacks By Persona

The Non-Technical Founder. Lovable for web, Vibecode if the product lives on a phone — it now ships real React Native apps with an App Store submission wizard. Plain-English chat to a full app either way. Lovable hit $100M ARR in its first eight months and is now approaching $600M — that's product-market fit with this exact persona. (If you want to try Vibecode: referral link.) Add v0 when the UI starts feeling rough. Skip Cursor and Claude Code for now. Real upside, wrong learning curve for a v1.

The Developer Who Wants To Ship Faster. Cursor. .cursorrules, codebase-aware chat, Agent mode — exactly the loops where devs lose time. Add Claude Code for the "do the whole thing" tasks: scaffolding services, writing migrations, analyzing unfamiliar repos. Skip browser builders. You'll hit the ceiling on day one.

The Weekend Warrior. Bolt.new. Zero setup, free tier holds for a Saturday, export if it turns into something. Add v0 for any UI-heavy bit. Skip anything with onboarding friction. Saturday isn't the day for API key setup.

The Enterprise Team. Copilot Business or Enterprise. Maybe Cursor Teams for the heavy-feature crew. Compliance, SSO, audit logs — that's what gets you past the CTO. Your team is already using AI tools (92% daily). Question is whether you're managing it or hoping nobody pasted prod code into a free chatbot. Skip consumer-tier plans for proprietary work.

The Hybrid Builder. Cursor for code-heavy work, Lovable or Replit for the quick-turnaround visual stuff. Don't commit fully to either world. v0 for components. Wispr Flow if you think while you talk.

---

Hosting: The Decision Most People Make Last

It should be near-first. The hosting story determines whether you own what you build or whether you're a tenant.

Fully managed. Lovable, Replit, Base44. Zero deployment overhead. Your app's availability and data live on their infrastructure. Understand the export story before you build something critical.

Bring-your-own. Cursor, Claude Code, Windsurf (now Devin Desktop). Setup overhead, full portability. Standard pairings: Vercel for frontend, Railway or Render for backend, Supabase for DB and auth.

Hybrid. Bolt (built-in hosting is now the default, Netlify one click away), v0 (native Vercel deploys plus backend services via its Connect panel), and some Vibecode setups. Ship today on their infrastructure, export when it earns it.

The question to ask out loud: if this platform disappeared tomorrow, what happens to my app and my data? If the answer scares you, that's information.

---

The Story

A friend runs a small events agency. No technical background. Needs a client portal — login, event timeline, vendor approvals, messaging.

She walks through the four questions before opening any tool. Technical: none. Duration: MVP, might become permanent. Building: web app with auth and DB. Budget: $30/mo.

Flowchart points to Lovable. One Saturday afternoon, she describes the portal in chat. By dinner she has something to show a real client. Not polished — the mobile layout's rough, one approval flow misfires — but real enough to test with a human.

The client has feedback in five minutes: they want to download the timeline as a PDF to send their venue. She types that into Lovable. It ships the feature. That loop — build, test, learn, extend — is the whole point.

Two months in, the portal is live with six paying clients. She's paying $25/mo for Lovable Pro. Hasn't written a line of code. When she eventually wants a feature Lovable's environment can't handle gracefully, she'll know exactly what that signal means: export, hand to a dev with Cursor, decide where it lives long-term.

Framework working correctly. Start in the right tool. Watch for the exit signal. Upgrade when the project earns it.

---

Where It Bites

Picking based on best case, not first week. Platform marketing shows you the finish line. The framework above is trying to show you the first mile. Optimize accordingly.

Treating the decision as bigger than it is. You can migrate. You can export. You can rebuild faster the second time because you understand what you're building. Three weeks choosing a platform is worse than three weeks building on the wrong one.

Over-engineering the initial stack. A non-technical founder with Cursor Pro, Railway, Supabase, and Vercel Pro before shipping a feature has a configuration problem, not a building problem. Match infra complexity to project complexity. Add layers as you earn them.

---

Gavin's Rule: Write Down Why You Picked It

Open a notes file before you write a single prompt. Answer the four questions. Write down which tool you chose and why.

Five minutes now. The payoff is six weeks later when a new platform launches and a friend swears it's the future. Your written criteria will tell you whether the new info actually changes anything — or whether you're just chasing novelty.

Stack stability is underrated. The builders who ship the most aren't using the best tools. They're the ones who went deep enough on one tool to learn its real edges.

---

Bottom Line

The landscape is wide enough that there genuinely is a right tool for your situation. The framework above gets you there without six weeks of comparison shopping. Four questions, one flowchart, one tool. Commit long enough to learn the texture. Revisit when the project has clearly outgrown it.

The decision today doesn't have to be the decision forever. It does have to be a decision, not an endless browse.

— Gavin