<!--
ROBOTS ON PAYROLL / THE VAULT
What this is: an opinionated decision guide for picking your first AI app builder (Lovable vs Bolt vs Replit vs Claude), with July 2026 prices.
How to use it: download this file and give it to your AI (Claude, ChatGPT,
Cursor, Lovable chat) as a reference. Say: "Use this as my playbook for picking my app builder."
Latest version + full guide: https://robotsonpayroll.com/resources/first-app-in-an-afternoon
-->

# The Tool Picker: which AI app builder to open today

Four tools can take you from a sentence to a working app this afternoon. They are more alike than their marketing admits, and the differences that matter are price, what happens when things break, and how the app gets online. Here is the honest version.

## The 30-second answer

**If you're not sure, start with Lovable.** Not because it's the "best" (that depends on what you're building), but because it's the hardest one for a beginner to get lost in: you type what you want, a preview appears on the right, and publishing is one button labeled Publish, top right. Every alternative below is a deliberate choice away from that default, and each row tells you when to make it.

If you want to spend $0 today and just prove to yourself you can build something, start in Claude instead. The free plan includes Artifacts (small apps Claude builds and runs right inside the chat window), and you can share one with a link.

## One-line verdicts

- **Lovable:** the best-looking output and the gentlest ride; heavy iterators hit the credit ceiling fastest.
- **Bolt:** the most generous free tier and it shows you the code as it works, which is either a feature or noise depending on your curiosity.
- **Replit:** the most complete package (builder, database, hosting, even a phone app), and the easiest place to spend more than you planned.
- **Claude:** free to try in the next 60 seconds, but Artifacts can't do user accounts or a real database; Claude Code (Anthropic's terminal-based builder, the pro tool) is where you graduate later, not where you start.

## The decision table

| What you're building | Open this | Why |
|---|---|---|
| A landing page, portfolio, or simple site | Lovable | Prettiest default output of the four, and Publish is one click |
| A web app with user accounts (habit tracker, client portal, booking tool) | Lovable | Its built-in Cloud backend wires up accounts and a database from a prompt |
| A calculator, quiz, game, or tool you want to share by link today, for $0 | Claude (free) | Artifacts build and run in the chat; hit Publish on the artifact to get a share link |
| Something where you want to watch and learn from the actual code | Bolt | Code editor and preview side by side; the other tools hide the code by default |
| An app you expect to keep growing, all in one place | Replit ($20/mo Core) | Database, hosting, and scheduled jobs live in the same tool, no third-party wiring |
| You already know you'll build a lot and want the pro path | Claude Pro ($20/mo) | Same subscription later covers Claude Code, so the money keeps working as you level up |

## Prices, July 2026

| Tool | Free tier | Paid entry | What paid gets you |
|---|---|---|---|
| Lovable | 5 credits/day, capped around 30/month | Pro, $25/mo | 100 monthly credits (roughly, 1 message that changes your app = 1 credit) |
| Bolt | 1M tokens/month, 300K/day cap | Pro, $25/mo | 10M tokens/month, no daily cap (tokens are the AI's unit of work; a feature costs a bundle of them) |
| Replit | Limited starter tier | Core, $20/mo | $25 of monthly credits; the Agent charges per task based on effort, simple edits usually under $0.25 |
| Claude | Artifacts included on the free plan | Pro, $20/mo | Higher usage limits, and the same plan covers Claude Code later |

Two pricing truths the landing pages soft-pedal:

- **Every tool meters you.** Credits, tokens, effort-based checkpoints: different names, same physics. Each message that changes your app costs something, which is why the prompts in this kit are written to get things right in fewer messages.
- **Replit is the only one where a single task has no visible price until it runs.** Effort-based pricing means a complex request can cost several dollars in one go. Fine once you know, startling the first time.

## The honest verdicts, one paragraph each

**Lovable** is the default for a reason: the interface is a chat plus a live preview, the designs it produces need the least fixing, and its Cloud backend means "add login" is a prompt, not a project. The catch is the meter. At 100 credits a month on Pro, a chatty back-and-forth afternoon can eat 30+ credits, so it rewards planning your prompt before sending (which is what the rest of this kit teaches anyway).

**Bolt** gives you the most free runway of the four, and the split view (code left, app right) quietly teaches you what a project actually looks like inside. The flip side: when something breaks, you're a bit more exposed to raw error text than in Lovable. If you're the kind of person who opens the hood at gas stations, pick Bolt and enjoy it.

**Replit** is the one your app is least likely to outgrow: the database, the hosting, and the always-on jobs are all native, and there's a genuinely good phone app for prompting your builder from the couch. It is also the most complex screen of the four and the least predictable bill. Better as a second tool than a first one, unless "all in one place" is exactly what you're after.

**Claude** is the zero-commitment proof that you can do this. Ask for "a tip-splitting calculator with a slider" in a free account and you'll have a working, shareable artifact in under two minutes. The ceiling is real, though: no user accounts, no stored data between visits, one screen. When you hit that ceiling, come back to this table and pick a builder. And file away the name Claude Code: it's the terminal tool serious builders use, it's covered by the same Pro plan, and it's the graduation path at the end of this kit, not today's job.

## Wrong reasons to pick a tool

- **"I saw a demo video where X built a whole startup."** Demo apps are chosen because they demo well. All four tools can build your first app.
- **"X uses the best model."** They all sit on frontier models and swap them constantly. The tool's workflow matters more than its model badge.
- **"I should learn the hard one first."** No. Ship something today in the easy one. The hard ones will still be there, and you'll arrive knowing what you actually need.

## When to switch

Switching later is cheaper than you think: your real asset after today is the description of your app (see the kickoff prompt in this kit), and that pastes into any of these tools in 30 seconds. Pick one, build, and revisit this file the day you feel the walls.
