From Family Bet to Live App: Building ActiveRivalry With Lovable

We had an idea called ActiveRivalry — an app where people stake money on real-world physical challenges. We built it end-to-end, with multi-tenant auth, per-user data, and a Stripe key managed cleanly, without opening the code once. Along the way, Lovable also cut our 25-hour WordPress builds down to about 5. Here's the honest version of how that went.

The idea came from inside our own team: someone at Daily Drop AI wanted to run a fitness challenge with their family — with real money on the line to keep everyone honest. It sounded like a product. It also sounded like the perfect test: could Lovable build a real app, and would that app be something people actually use? Two birds, one build.

ActiveRivalry came first: we prompted our way to a working money-stake challenge app — auth, personalized data per user, even a logo (weak, but it tried). The visual editor did the heavy lifting whenever the AI's output was almost-right: click the element, change exactly that, upload an image directly.

ActiveRivalry is live at activerivalry.com — and somewhere along the way it stopped being a test project. We liked using it enough to keep it. We're running it as an honest experiment: private rivalries with friends first, working toward a real launch. Real-money stakes, transparent fees kept as low as we can (they keep the lights on), and winners split the pot. Go see what a no-code build can ship.

The test project earned Lovable a permanent seat. Client marketing websites came next: a 5–15 page site that takes us ~25 hours in WordPress/Elementor took ~5 in Lovable — we topped up credits twice in one month pushing one as far as it would go. We even built a small internal iOS tool in about five prompts. Nothing we tested failed outright; when errors appeared, Lovable often caught and fixed its own.

New to Lovable? Our scored review covers what it is, the pricing reality, and the full pros and cons.

Our one rule for AI app builders: don't ship AI slop — the cards and components you can instantly tell were generated. Our fix is pairing Lovable with 21st.dev: pull component inspiration and copy prompt-ready designs, then focus your Lovable prompts on UX/UI improvements. The difference is immediate. Grab designs from 21st.dev and paste the prompt straight in.

Three habits that made Lovable dramatically better for us: (1) Plan first, even for small changes — let the AI study the codebase before it builds. (2) Let a terminal agent write your prompts: we connect Claude to Lovable's official MCP server for gap-finding and higher-quality prompts — with one guard: make it show you each prompt before sending, because prompts are credits. One caveat: the MCP ignores plan mode and builds immediately; the workaround is drafting the prompt outside and pasting it in yourself. (3) Dictate — we use Wispr Flow. Long, specific prompts beat short vague ones, and talking is faster than typing.

We ran the same prompt through Lovable, Bolt.new, and Replit at a similar price point. Lovable won our testing — better first output, and the visual editor sealed it.

The roadmap is deliberately simple: figure out the cleanest way to verify winners — including whether we can integrate with systems that can verify results for us — and expand the kinds of competitions you can run. We'll keep building it in Lovable until it gives us a reason not to.

Lovable lets you remix any project: duplicate it, rebrand it, and ship it again. Build a simple CRM for one industry, then remix it for the next client or niche. For agencies and template sellers, that's a genuine business model hiding in a feature.

Lovable scored an 8.8/10 in our full review — a permanent spot in our stack for simple-to-mid apps and websites. It's the tool we reach for first when we want something real online this week.