Know What AI Builders Can and Cannot Do

Why this matters

The fastest way to quit on day one is to pick a project the tool cannot actually build yet, watch it fail, and conclude the whole approach is hype. AI builders like Lovable and Bolt are genuinely great at a specific shape of app — web apps with forms, lists, dashboards, and simple logic — and genuinely bad at others, like a native iPhone app in the App Store or a live multiplayer game. Knowing which side of that line your idea sits on before you start is the difference between shipping something real this afternoon and spending three frustrated hours fighting the tool. Calibrating this now also lets you scope your first build so it is ambitious enough to feel like a real product but small enough that you can finish it today.

Demo

Here is the honest capability map as of 2026. IN RANGE and easy today: a habit tracker, a link-in-bio page, an expense splitter, a wedding RSVP form with a guest list, a small business landing page with a contact form, a job-board directory, a booking/appointment picker, an internal dashboard that shows numbers from a spreadsheet, a simple flashcard study app. These are web apps, single database, forms in and lists out. IN RANGE but harder (expect more iteration): anything that charges money (Stripe checkout), anything with user login and private accounts, anything pulling live data from another service's API. OUT OF RANGE today: a native mobile app you download from the App Store (these tools build web apps that work on a phone browser, not true iOS/Android apps), a real-time multiplayer game, video editing or heavy image processing, anything needing millisecond-precise timing, and anything at the scale of Uber or Instagram. To make this concrete, sort these ten ideas: (1) a tip calculator — YES, easy. (2) a personal book tracker — YES, easy. (3) a Tinder-style native dating app on the App Store — NO, native mobile. (4) a neighborhood tool-lending directory with listings and a request form — YES. (5) a live 8-player online chess site — NO, real-time multiplayer. (6) a freelancer invoice generator that emails a PDF — MOSTLY, harder (email + PDF). (7) a wedding website with RSVP — YES, easy. (8) a Spotify competitor streaming music — NO, scale and licensing. (9) an internal dashboard reading a Google Sheet of sales — YES. (10) a paid membership site with monthly billing — HARDER, needs Stripe and login. Notice the pattern: forms, lists, and simple math are yes; native mobile, real-time, and massive scale are no.

Try it yourself

1. Write down three app ideas you personally care about, then label each one IN RANGE, HARDER, or OUT OF RANGE using the demo's categories — and write one sentence explaining the label.
2. For any idea you labeled OUT OF RANGE, rewrite it as a smaller web-app version that IS in range (e.g. 'native fitness app' becomes 'web workout logger I open in my phone browser').
3. List every reason your first-choice idea might be HARDER (money, login, live data from another app) so you can dodge those for your very first build.
4. Pick the single idea that is most 'ambitious but finishable today' and write it at the top of a notes file — this becomes your candidate for the module project.

References

• Lovable docs — what you can build
• Bolt.new — supported frameworks and limits
• MDN — What is a web app? (progressive web apps intro)