I rebuilt Lovable's home screen and changed one component, the one I think has the biggest effect on revenue. Try it, then flip back to the original.
Independent concept by Sascha Thiele. Not affiliated with or endorsed by Lovable.
Send a few prompts and watch the sidebar. Each message costs one of the 5 free daily credits.
Best on a larger screen. On a phone, tap the cards below the prototype to see each state.
Upgrade to keep building. This is the first time you see your balance, and it's already zero.
Type into the prompt and press Enter, or use "Send a prompt for me". Or click a card below to jump straight to that state.
"Upgrade to Pro" becomes a live meter. It counts in edits, not abstract credits, and always shows when credits reset.
Show in prototype ↑A small "≈ 1 credit" label next to Build. You see the cost before you spend, so the bill never surprises you.
Show in prototype ↑With one credit left, mid-build, the meter turns into an offer. It's sized to your own pace, not "unlock more features".
Show in prototype ↑No blocking wall. The prompt is kept, the path forward sits inline, and "remind me" brings you back when credits reset.
Show in prototype ↑I went through seven screens: home, ⌘K, connectors, plans, referrals, growth and bonuses. Four things stood out.
Upgrade to Pro · Unlock more features
sits in the corner all day, the same for every user at every moment. When a nudge never changes, people learn to stop seeing it.
Credits drive the business, but the home screen never shows them. The first time you meet them is at the wall, mid-idea. That is the worst moment to show a paywall, and it's the only time one appears.
Pro and Business both default to 100 monthly credits
. Users who have never seen their own burn rate can't tell which tier fits, so they pick the smallest one.
Three different modals are about getting more out of Lovable. All of them are worth more when people can see the thing they earn: credits.
Turn the static "Upgrade to Pro" card into a live credit meter. Show the balance, show the cost before sending, and offer the upgrade when intent peaks, at a tier that fits the user's pace.
The defaults are estimates built from public figures (source below), plus two assumptions of my own, which are marked. You know the real numbers. Move the sliders and the result updates.
Illustrative only. It's simple cohort math with linear retention decay and no reactivation or seasonality, in USD. The real number comes from the A/B test below.
Source for users, subscribers and ARR: DemandSage, Lovable Statistics (2026), which compiles TechCrunch, Sacra and Dealroom. Lift and retention are my assumptions.
I'm Sascha, a freelance design engineer in Leipzig. I spent six years designing products for Mercedes-Benz, Amazon, Ottobock and Deutsche Bahn. Each role moved me closer to implementation. Since 2024 I've built the things I design myself, including two iOS apps that are live on the App Store.
I just finished a full-time AI engineering programme, so when I design for AI products I know how they work underneath.
A European tech job board where every listing shows its salary and paid ranking is always labelled. Candidates, employers and moderators each get their own interface, and I designed and built all three.
An app blocker for ADHD brains. You write down one task, and the apps you picked stay paused until you say it's done. A paused feed shows your task instead. No countdown, no streaks, no guilt.
A habit tracker built around widgets. Check in from the Home or Lock Screen, ask Siri for your streak, share a Year in Review. Free, with no ads, no account and no subscription.
A native iOS app for learning Japanese, with reading, vocabulary and review in one place. A Python pipeline gives every text correct furigana. Monetised with StoreKit 2.
An agentic RAG reading assistant for Japanese and Hebrew learners. It answers vocabulary questions and simplifies any text to a target level.
UX/UI concepts, interface design and design systems. AI/ML prototypes from data pipeline to interface. Consulting clients and stakeholders.
Concept, UX/UI and development end to end. Two apps live on the App Store, and Hane in beta with 200+ TestFlight users.
Ran live auctions and led the auction house's digital transformation, website and print design.
App concept and design for large brands, including Mercedes-Benz, Amazon, Deutsche Bahn, Samsung and Tagesschau. Stakeholder consulting and design management.
Concept and design of apps and websites, client consulting, design management.
App and website interfaces, motion design, social content.
Six years of agency and consulting work.
How the designs get built.
From the AI engineering programme and Readr.
Why LovableGiving people tools that let them create things they could never have built before is one of the most powerful things software can do. That's what Lovable does, and I want to help build it.
As a design engineer I sit exactly where Lovable aims: I think in interfaces, I can read the code, and I want to get from idea to working product without waiting on anyone. I'd be designing the product I'd want to use myself.
This application is built around a revenue hypothesis, not a mood board. That's also how I worked with enterprise clients: every screen had to earn its place.
This page is one hand-written HTML file. My products are shipped code, not Paper files. Engineers get decisions they can build from, and I can build them too.
I'd love to talk about this change, the ones I left out, and what I could build at Lovable.