By the PrototypeVerdict Test Team · Tested: Oct 29, 2026
Most “vibe coding examples” online are success stories picked after the fact. This page is the opposite: three ordinary apps we built in every tool we test on Oct 29, 2026, sorted by how well they went for a beginner. A landing page worked everywhere. A tracker worked with a few catches. A portal with logins worked, but needed care. Then there’s a list of things we haven’t tested, so you know where the evidence stops.
Quick verdict
Our verdict
| Works well today | One-page sites with a form: working in about 2 to 5 minutes, 0 fixes, in Bolt, Lovable and Base44 |
| Works with care | Trackers and simple business tools: worked first time, but check where data is saved and how rules are applied |
| Possible but fiddly | Apps with logins and an admin: needed a fix or a manual step in two of three AI builders |
| Not tested by us | Payments, app-store mobile apps, high traffic, regulated or sensitive data |
| Rule of thumb | If it’s one screen, one action and one kind of saved data, a beginner can build it today |
Works well today: landing pages and simple sites
What we built: a one-page site for a made-up bookkeeping service, with a headline and button, three features, a three-tier pricing table, three sample testimonials, five FAQs and a contact form that saves submissions.
| Tool (plan) | Time to working | Fixes | Notes |
|---|---|---|---|
| Bolt (Free) | ~2–3 min | 0 | ~15K tokens, about 5% of a free day; showed a stray “\u2019” |
| Lovable (Pro) | ~2 min 50 s | 0 | 1.5 credits; invented its own FAQ and plan copy |
| Base44 (Builder) | ~5 min | 0 | ~1.65 message credits |
| Softr (Basic) | ~4 min to ready | 0 | Added a default yearly price toggle and a “Trusted by 1,000+” badge we didn’t ask for |
Good for: a page for a side business, an event sign-up, a waitlist, a portfolio. Catch: you must proofread. Every tool wrote copy we didn’t supply. Lovable also said email notifications for form submissions need a paid plan and your own domain, so check how you’ll actually see new messages.
We built our landing page on Bolt’s free plan too, so you can try one at bolt.new without paying.
Works with care: trackers and simple business tools
What we built: an invoice tracker with clients, invoices, line items and automatic totals, Draft/Sent/Paid/Overdue statuses, and a dashboard.
| Tool (plan) | Time to working | Fixes | Catch |
|---|---|---|---|
| Bolt (Free) | ~9 min | 0 | ~100K tokens, a third of a free day |
| Lovable (Pro) | ~4 min | 0 | Saved data in the browser only, until asked for a database |
| Base44 (Builder) | ~12.5 min | 0 (one “continue”) | “Overdue” applied only to invoices marked Sent |
| Softr (Basic) | Didn’t fully work | 4 follow-ups | No Overdue status; couldn’t add invoices or line items in the app preview |
Good for: trackers, simple CRMs, inventories, sign-up lists, internal checklists. Catch: rules get interpreted narrowly, and “saved” doesn’t always mean “in a database.” Test each rule with real-looking (fake) data.
Possible but fiddly: apps with logins, roles and uploads
What we built: a client intake portal for a small accounting firm. Clients sign up, fill in a form, upload documents and see their own submissions; an admin sees everything and changes statuses.
| Tool (plan) | Time to working | Fixes | What it took |
|---|---|---|---|
| Base44 (Builder) | ~4.5 min | 0 | Passed our check with a separate client account |
| Lovable (Pro) | ~9 min | 1 | A message to create the first admin; a blank page after email confirmation; passed the separate-client check |
| Bolt (Free) | ~9 min | 1 | A login fix (~84K tokens) and a manual admin edit in the database; separate-client check not run |
| Softr (Basic) | Didn’t work | 1 | Plan limits switched off its user groups, so a test client could see every submission |
Good for: a prototype portal you test with fake data. Catch: privacy is the whole point of a portal, and it’s the part most likely to go wrong. Always test with a second account. Step-by-step: build a client portal with AI and how to add login and a database.
Step-by-step builds of this portal: our Lovable tutorial for beginners and Base44 tutorial for beginners.
Where no-code tools fit
Softr, a no-code builder with AI help, built a real database for all three apps on its Basic plan, and its landing page worked. But its portal and invoice tracker didn’t fully work on Basic: user groups were switched off by a plan limit, and invoices couldn’t be added in the preview. Softr recommends Pro for client-facing apps, which we haven’t tested. Not scored yet: our tests need Softr Pro. See our Softr notes.
Not tested by us yet
We’d rather tell you where our evidence stops:
- Payments. None of our test apps took money. A wrong payments setup risks real money, so get help or wait for our test.
- App-store mobile apps. All our apps were web apps that worked at phone width. Bolt documents an Expo integration for iOS and Android; we haven’t tried it.
- Publishing and real traffic. Nothing we built was published. See how to publish your AI-built app for the docs-based steps.
- Sensitive or regulated data. Tax records, health data, card numbers: don’t put them in a prototype. Our security scans flagged dependency issues in Lovable (12) and Base44 (7 high-severity).
- Email notifications, integrations and AI features inside your app. Not part of our tests.
- Big apps. Our biggest app had a handful of screens. On Bolt, token cost grows with project size.
10 starter ideas by job
Only ideas 1–3 and 10 are things we’ve built (10 is the same landing page, written up for marketers). The others are ideas, not tested, matched to the closest app we did test so you know roughly what to expect.
| # | Idea | For | Closest tested build | Status |
|---|---|---|---|---|
| 1 | Landing page with a contact form | Anyone with a service | Landing page | Tested |
| 2 | Invoice tracker | Freelancers | Invoice tracker | Tested |
| 3 | Client intake portal | Accountants, agencies | Client portal | Tested |
| 4 | Document request checklist | Accountants | Client portal | Not tested (prompt on /prompts/) |
| 5 | Client deadline tracker | Accountants | Invoice tracker | Not tested (prompt on /prompts/) |
| 6 | Engagement letter drafter (internal) | Accountants | Invoice tracker | Not tested (prompt on /prompts/) |
| 7 | Client onboarding form | Freelancers | Client portal | Not tested (prompt on /prompts/) |
| 8 | Time tracker | Freelancers | Invoice tracker | Not tested (prompt on /prompts/) |
| 9 | Portfolio page | Freelancers | Landing page | Not tested (prompt on /prompts/) |
| 10 | Marketing landing page | Marketers | Landing page | Tested (same build as #1; see vibe coding for marketers) |
Prompts for ideas 1–9 are in our prompt library. Role guides: vibe coding for accountants and vibe coding for marketers.
From idea to tool
Pick the smallest version of your idea, then choose a tool: best AI app builders for beginners. In short: Base44 gave us the fewest problems, Lovable is the pick if you want the code, and Bolt let us build all three apps without paying. Before you begin, read vibe coding for beginners.
A note on cost: our three apps used ~300K Bolt tokens (the whole free day), 5.5 Lovable credits from the balance plus 5 daily credits, and ~8.75 Base44 message credits. We quote Lovable’s balance because Lovable’s credit counter was inconsistent in our test: the per-project usage page came to about 10.2 for the same three apps.
Related tests: Build a client portal with AI · Lovable review · Base44 review
Then write your first prompt with how to write prompts for AI app builders, and if you want to spend $0, see how to build an app with AI for free.
What can you build with vibe coding FAQ
What can a beginner build with vibe coding?
From our tests: landing pages with working forms (easy), trackers like an invoice tracker (worked, with catches), and portals with logins and an admin (worked in Lovable, Base44 and Bolt, with more care). All were web apps tested with fake data.
Can you really build an app with AI?
Yes, for small web apps. On Oct 29, 2026, three AI builders each produced a working landing page, invoice tracker and client portal from one prompt each, with 0 to 1 bug-fix messages per app.
What are some vibe coding examples?
Our three test apps: a bookkeeping landing page, a freelancer invoice tracker and an accounting-firm client portal. Results for each tool are on this page, and the prompts are in our prompt library.
Can I build a mobile app with vibe coding?
Our apps were web apps that worked at phone width. Bolt documents an Expo integration for iOS and Android, and Lovable suggests installable web apps, but we haven’t tested any app-store path.
Can I build an app that takes payments?
We haven’t tested payments in any tool yet. Because mistakes can cost real money, we’d suggest getting help, or waiting for our test, before building one.
What’s too hard for vibe coding right now?
In our tests, the fiddly part was logins and who-sees-what. Beyond that, we haven’t tested payments, app-store apps, big apps or sensitive data, so we can’t vouch for them.
Update history
- October 29, 2026: First published. Results from our Oct 29, 2026 test round (Bolt Free, Lovable Pro, Base44 Builder, Softr Basic).