Lovable

Huge update: GPT-5 is here 🔥 We just integrated it into Lovable Here is what's new: → System-level thinking. → Multi-component builds. Auth systems, payment flows, etc. → Business logic that works. GPT-4: "Build me a todo app" GPT-5: "Build me a SaaS with user auth, payments, admin dashboard, and email automation" We're not improving code generation. We're eliminating the need to code. GPT-5 preview starts 11AM PT today. Ends Sunday midnight. The gap between idea and working product just disappeared. Go try and build something!!

OpenAI's GPT-5 is fresh out of the gates, and is now in Lovable. As foundation models advance, the scope and complexity of what's possible to build in Lovable continues to expand. → Better reasoning. → Logic that works. → Increasing possibility. GPT-5 Preview starts now, ending Sunday midnight.

OpenAI's GPT-5 is out, let's see it in action in Lovable: GPT-3 launched in 2020 and, at the time, most people missed the bigger picture. In 2023, GPT-4 dropped with a significant improvement on code generation – and it led to GPT-Engineer, the first version of Lovable. The capabilities of GPT-4 were incredible at the time. But with the latest advancements in foundational models, the scope and complexity of what’s possible to build has expanded dramatically. In the short time that Lovable has existed, people have built serious software businesses end-to-end with Lovable, generating millions of dollars in revenue. We're looking forward to seeing more successful businesses get built on Lovable. Today by 11AM PT (within one hour from now), we'll be making GPT-5 available in limited preview until midnight PT on Sunday August, 10th. Try it and let us know what you think: lovable.dev

Announcing the winner of Lovable Shipped, taking home $100k. Over the past 6 weeks, the 5,800 contestants have taken raw ideas and turned them into real products. Today, Mahmoud Halat is taking home the grand prize of $100k USD with his product GiveFeedback, which turns scattered client feedback into clear next steps (built 100% on Lovable in 6 weeks!) Our judges were impressed by Mahmoud’s understanding of the problem, his product’s potential to scale, and the clarity of his pitch. Congrats to Mahmoud and the GiveFeedback team for making feedback lovable ❤️

lovable shipped - demo day our top 10 finalists from the 6 week lovable shipped program pitch live before a panel of vc judges for their chance at $100k grant. christian peverelli of wearenocode joins founder of lovable anton osika to discuss the future of no code and how to make something lovable. guest panel of judges include nikita andersson (hummingbird), jeff becker (antler) and bjarke klinge staun (ex-creandum).

After 100 hours of prompting AI for UI design, I discovered most designers are doing it completely wrong. The difference between getting generic templates vs. production-ready components comes down to 10 specific techniques. Here's what I learned: 𝗟𝗲𝘀𝘀𝗼𝗻 𝟭: 𝗦𝗽𝗲𝗰𝗶𝗳𝗶𝗰𝗶𝘁𝘆 𝗯𝗲𝗮𝘁𝘀 𝗰𝗿𝗲𝗮𝘁𝗶𝘃𝗶𝘁𝘆 Bad prompt: "Design a dashboard" Good prompt: "SaaS analytics dashboard, dark theme, 3-column layout, card-based metrics, Tailwind CSS" The AI needs constraints to be creative. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟮: 𝗦𝘁𝗲𝗮𝗹 𝗹𝗶𝗸𝗲 𝗮𝗻 𝗮𝗿𝘁𝗶𝘀𝘁 (𝗹𝗶𝘁𝗲𝗿𝗮𝗹𝗹𝘆) Don't ask AI to "be creative." Reference real products: "Design like Linear's issue view but for customer support tickets" AI understands patterns better than aesthetics. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟯: 𝗖𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁𝘀 > 𝗳𝘂𝗹𝗹 𝗽𝗮𝗴𝗲𝘀 Start with: "Create a pricing card component" Not: "Build me a pricing page" 𝗟𝗲𝘀𝘀𝗼𝗻 𝟰: 𝗧𝗵𝗲 𝗯𝘂𝘇𝘇𝘄𝗼𝗿𝗱 𝗵𝗮𝗰𝗸 Generic words = generic results. Specific buzzwords = sharp designs. Instead of "modern": use "glassmorphism", "brutalist", "neubrutalism" Instead of "clean": use "minimal", "Swiss design", "monospace" Buzzwords are design shortcuts. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟱: 𝗔𝗹𝘄𝗮𝘆𝘀 𝗶𝗻𝗰𝗹𝘂𝗱𝗲 𝘁𝗵𝗲 𝘁𝗲𝗰𝗵 𝘀𝘁𝗮𝗰𝗸 "React component with Tailwind CSS" hits different than just "component." AI knows framework patterns. It'll give you proper JSX, class names, and structure. The more technical context, the better the output. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟲: 𝗜𝘁𝗲𝗿𝗮𝘁𝗲 𝗶𝗻 𝗽𝘂𝗯𝗹𝗶𝗰 (𝘄𝗶𝘁𝗵 𝗔𝗜) Don't hide in private chats. Build in Lovable. Share the process. Real feedback > perfect prompts. Users care about function over form anyway. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟳: 𝗠𝗼𝗯𝗶𝗹𝗲-𝗳𝗶𝗿𝘀𝘁 𝗽𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴 "Mobile-first responsive" isn't enough. Be explicit: "Mobile: single column, touch targets 44px minimum" "Desktop: 3-column grid, hover states" Design for thumbs first. Scale up second. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟴: 𝗧𝗵𝗲 𝟴𝟬/𝟮𝟬 𝗽𝗿𝗼𝗺𝗽𝘁 𝗿𝘂𝗹𝗲: 80% of results come from these 4 elements: Component type Visual style Tech stack Layout structure Master the basics first. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟵: 𝗔𝗜 𝗰𝗮𝗻'𝘁 𝗿𝗲𝗮𝗱 𝘆𝗼𝘂𝗿 𝗺𝗶𝗻𝗱 "Make it better" = vague feedback. "Increase contrast, larger text, more padding" = actionable. Treat AI like a junior designer. Be specific, patient, and iterative. 𝗟𝗲𝘀𝘀𝗼𝗻 𝟭𝟬: 𝗟𝗮𝘆𝗲𝗿 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗺𝗽𝘁𝘀 Don't try to get perfect results in one shot. First prompt: "Create a login form component" Second prompt: "Add error states and loading animations" Third prompt: "Make it mobile-responsive with proper spacing" Build complexity step by step. AI works better with focused, layered requests than one-shot prompts. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹𝗶𝘁𝘆: 𝗔𝗜 𝗶𝘀𝗻'𝘁 𝗿𝗲𝗽𝗹𝗮𝗰𝗶𝗻𝗴 𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗿𝘀. It’s just exposing who can’t think in systems, structure, and intent. Which lesson hit hardest? Comment the number (1-10) and I'll share my best prompt template for that specific use case.

We're live coding this Wednesday with our friends Lovable and Trigger.dev! Come learn how to build an uptime monitoring service from scratch using agentic workflows. Our uptime monitor will have the follow features: ✓ Create and manage uptime monitors (URLs and frequency) ✓ Start/stop monitors on demand ✓ Home dashboard with global stats across monitors ✓ Drill-down pages for individual monitors with detailed and historical monitoring stats ✓ Time filtering across all views ✓ A public Status Page per customer ✓ Email notifications for downtime We'll use Tinybird Code to build our analytics backend, Lovable to build our frontend, and Trigger to handle orchestration and background tasks. Link to register is in the comments.

We’re giving away $100K to one of the 10 finalists in Lovable Shipped. Over the past 6 weeks, they’ve taken raw ideas and turned them into real revenue-generating products. Now, it all comes down to one moment: They're pitching live (online) on Wednesday August 6th at 6:30PM CET. Whether you’re an investor looking for the best Lovable startups or someone wanting to build your own startup with Lovable, you should watch it live! RSVP here: https://lu.ma/ype8fy6l

We're live coding this Wednesday with our friends Lovable and Trigger.dev! Come learn how to build an uptime monitoring service from scratch using agentic workflows. Our uptime monitor will have the follow features: ✓ Create and manage uptime monitors (URLs and frequency) ✓ Start/stop monitors on demand ✓ Home dashboard with global stats across monitors ✓ Drill-down pages for individual monitors with detailed and historical monitoring stats ✓ Time filtering across all views ✓ A public Status Page per customer ✓ Email notifications for downtime We'll use Tinybird Code to build our analytics backend, Lovable to build our frontend, and Trigger to handle orchestration and background tasks. Link to register is in the comments.