Blitz lets you deploy sites with full backend (auth + db + api) from literally anywhere.
Just ask "deploy to https://t.co/91DgPrbIYu". No signup needed.
I asked Claude Cowork to make a dashboard showing my 2026 claude code usage. 16B tokens, $36k API-equivalent so far ๐
@BorisMPower You guys should really enable network egress settings in ChatGPT so people can use any API based app builder. https://t.co/93b89TlOD2 has it
@minjunesh@spencerc99 for real-time persistence of mouse movement data 8. Each visitor sees the 'ghost' of the last person who browsed the page, creating a collective portrait effect
@minjunesh@spencerc99 previous visitor's complete mouse path 4. Render the cursor as a colored dot 5. Paint the entire mouse path as if the mouse button was held down with a paintbrush effect 6. Create a continuous trail/stroke following the path the previous visitor took 7. Use WebSockets or similar
You can make websites directly from Claude Design
Go to Settings > Capabilities > allow network egress and set domain allowlist to "All domains"
Then in chat just say "deploy my design to https://t.co/91DgPrbIYu, fetch the website first to understand"
@minjunesh@steveruizok UI that could connect to a blog-focused index or aggregator API. For the MVP, implement the frontend with mock data representing diverse personal blog posts, with the architecture ready to plug into a real search backend.
@minjunesh@steveruizok bloggers, not corporate blogs or news sites), 3) Results should show blog post title, author/blog name, snippet, and date, 4) Filter options for date range and possibly topics/tags, 5) The aesthetic should feel indie/personal - not corporate. This should be a functional search
@minjunesh@VikParuchuri processing, allowing users to upload PDFs and see OCR results. Reference images show the NYT moon landing front page (example of complex document OCR) and an olmOCR-bench performance chart comparing various OCR models.
@minjunesh@VikParuchuri Built and deployed it: https://t.co/EyuvkBLq1Y
Built a client-side browser-based OCR demo using the Surya OCR 2 model (650M params) that can process real PDF files. The demo should run the actual model inference in the browser (WebAssembly/WebGPU) without server-side
@minjunesh@VikParuchuri Built and deployed it: https://t.co/WLghs8CwuI
Built a browser-based OCR demo app that showcases Surya OCR 2 capabilities. The app should: 1) Allow users to upload a PDF file, 2) Display the uploaded PDF pages visually, 3) Demonstrate OCR text extraction workflow with a test
@minjunesh@steveruizok blog post titles, snippets, author names, and blog URLs. Style it with a warm, indie-web aesthetic that feels different from corporate search engines. Could include curated categories like tech, life, creative writing, etc.
@minjunesh@steveruizok Built and deployed it: https://t.co/LCBuQEc7xZ
A search engine specifically for personal blog posts. Features: a clean search interface where users can search across personal blogs/indie web content. Include a search bar, filters (by date, topic), and display results showing
@divamgupta@PrismML prompt input field, a generate button, and display the generated image. The UI should explain that inference runs locally on the user's hardware. Use WebGPU or WASM-based inference if the model weights/runtime are available, or create a compelling demo UI that simulates the
@minjunesh@quarqlabs Include a clean header explaining this is a feature summary. Since the article content isn't directly accessible, create a placeholder gallery structure that can showcase multiple features in an attractive grid or carousel layout with sample feature cards demonstrating the UI