Spent a few hours fully automating the PaperFox.ai mobile app release workflow. Now a single command packages and uploads the new version for both iOS and Android. It took a fair bit of manual setup up front, but every release after this is so easy. I also asked Claude Code to draw a flowchart from my coding session. I tried both Excalidraw and Mermaid — the Excalidraw version is better laid out and just looks nicer. #BuildInPublic #MobileDev #ReactNative #DevOps #Automation #ExpoEAS #iOS #Android #ClaudeCode #IndieHacker #AIBuilders #PaperFox

Spent a few hours fully automating the PaperFox.ai mobile app release workflow. Now a single command packages and uploads the new version for both iOS and Android. It took a fair bit of manual setup up front, but every release after this is so easy. I also asked Claude Code to draw a flowchart from my coding session. I tried both Excalidraw and Mermaid — the Excalidraw version is better laid out and just looks nicer. #BuildInPublic #MobileDev #ReactNative #DevOps #Automation #ExpoEAS #iOS #Android #ClaudeCode #IndieHacker #AIBuilders #PaperFox

Another example of what AI won't get right unless you supervise it. I asked Claude Code to build the AI matching algorithm for registrants (paperfox.ai/blog/registratio…) as an async job on Inngest. How it went: 1. It set max output tokens to 3000 → worked fine locally on a small sample, then errored out mid-run in production once real data hit it. 2. I asked it to bump the limit to 20000 and surface a clean error if exceeded → it worked, but the 21 batches ran one after another. Total time: ~10 minutes (screenshot 1). Functionally correct — just sequential for no good reason. 3. So I told CC: read the Inngest docs, find the best practice for fan-out, and optimize. Second pass (screenshot 2): all 21 batches kicked off in parallel. Same workload, finished in under 3 minutes — ~4x faster. The model can write the code, pick a token limit, and call the API just fine. What it won't do on its own is ask "is this actually the right shape for the job?" That part is still on you. This is what we need to teach our students too. #AI #ClaudeCode #Inngest #SoftwareEngineering #PaperFox #LLM #BuildInPublic #AIBuilders #AgenticEngineering

How do you manage your AI coding agent's memory (important part of context engineering)? Most small projects don't run into this issue, but we keep adding rules to CLAUDE.md as we ship features and fix bugs on PaperFox.ai. The file regularly grows past 40k characters, and Claude Code (CC) warns that loading this much every session tanks performance (see Screenshot 1). I used to just ask CC to condense the file by trimming details. That works, but lately we've been hitting the limit too often. So I asked CC to read the official docs. Turns out Claude Code has a feature for exactly this: .claude/rules/ — markdown files with frontmatter declaring which paths they apply to. They load 𝗼𝗻𝗹𝘆 when Claude touches matching code. I split our 600-line CLAUDE.md into a root 14 scoped files (see Screenshot 2). Then I built a small companion skill, /add-to-claude-md, that asks "is this rule cross-cutting or scoped?" before writing the entry — so the CLAUDE.md file doesn't bloat back up. Now our Claude Code memory is modular and context-scoped — much easier to extend and maintain. #AI #ClaudeCode #ContextEngineering #AgenticAI #AICodingAgent #PaperFox #BuildInPublic

How I used AI to automate a chore I don't like to do. Real business problem: we used AI for SEO and got PaperFox.ai ranked #1 on Google. But Baidu, the largest search engine in China, wasn't properly indexing us. Had to fix that. Getting indexed by Baidu is the kind of chore I hate: the console is complicated, full of forms, and I'd have to read a pile of docs just to figure out what to click and what to fill in. This is how I used Claude Code (CC) to do it: - 𝗥𝗲𝗮𝗱 𝗮𝗻𝗱 𝗻𝗮𝘃𝗶𝗴𝗮𝘁𝗲𝗱 𝘁𝗵𝗲 𝗦𝗘𝗢 𝗰𝗼𝗻𝘀𝗼𝗹𝗲. CC drove a Chrome tab through Baidu's signup, navigated the wizard, and guided me all the way. - 𝗥𝗲𝗮𝗱 𝘁𝗵𝗲 𝗱𝗼𝗰𝘀, 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗲𝗱 𝘁𝗵𝗲 𝗕𝗮𝗶𝗱𝘂 𝘃𝗲𝗿𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗰𝗼𝗱𝗲, 𝗮𝗻𝗱 𝘄𝗿𝗼𝘁𝗲 𝗶𝘁 𝗶𝗻𝘁𝗼 𝘁𝗵𝗲 𝗰𝗼𝗱𝗲𝗯𝗮𝘀𝗲 — meta tag straight into our root layout, commit and deploy. - 𝗣𝘂𝘀𝗵𝗲𝗱 𝗨𝗥𝗟𝘀 𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗕𝗮𝗶𝗱𝘂'𝘀 𝗔𝗣𝗜 instead of me clicking submit 11 times in a clunky web form (foreign-hosted sites get throttled and capped at a daily manual upload quota). - 𝗘𝗻𝗰𝗼𝗱𝗲𝗱 𝘁𝗵𝗲 𝗽𝗹𝗮𝘆𝗯𝗼𝗼𝗸 𝗮𝘀 𝗮 𝗿𝗲𝘂𝘀𝗮𝗯𝗹𝗲 𝗦𝗸𝗶𝗹𝗹 — 𝗜 𝗮𝘀𝗸𝗲𝗱 𝗖𝗖 𝘁𝗼 𝗱𝗼 𝘁𝗵𝗶𝘀: /baidu-push <conf-slug> for the next conference. - 𝗕𝘂𝗶𝗹𝘁 𝗮 𝗱𝗮𝗶𝗹𝘆 𝗿𝗼𝘂𝘁𝗶𝗻𝗲 𝘁𝗵𝗮𝘁 𝗮𝘂𝘁𝗼-𝗿𝘂𝗻𝘀 𝗶𝗻 𝘁𝗵𝗲 𝗰𝗹𝗼𝘂𝗱. CC assembled a URL queue, set up a remote agent to push 10/day until it drains (~14 days). Zero further attention from me. This workflow can be applied to many other tasks, and I think many people still don't know how to do this with AI. Hope this helps. #AI #ClaudeCode #SEO #Automation #AgenticAI #DeveloperTools #BrowserAutomation #PaperFox

We just shipped something I'm really excited about — PaperFox now supports MCP (Model Context Protocol), so you can talk to your conferences directly from Claude, ChatGPT, or any MCP-compatible AI assistant. Things you can just ask: - "How many papers were submitted to CSWIM2026?" - "Show me submission #42" - "What's the review progress for the AI track?" - "Which reviewers haven't finished yet?" - "When and where is the keynote for WISE2026?" The most fun part: paper submission via MCP Apps. The AI collects your details conversationally, then opens a pre-filled form right inside the chat window. Upload your PDF and submit without opening a browser. I believe the chat window is becoming the new operating system and the next shift in UI/UX is definitely conversational — you describe what you need, and the AI figures out which tools to call. The web app isn't going away — it's still the best interface for complex workflows. But for the dozens of quick questions chairs and authors ask every day, a conversation is just better. Still early and experimental. Try it and tell me what you think. paperfox.ai/blog/mcp-server #MCP #AI #BuildInPublic #PaperFox #AcademicConferences #LLM #Claude #ChatGPT #AIAgents #EdTech #SaaS #ConversationalAI

"Coding is largely solved" — If you haven't listened to this episode - highly recommend! I recently feel Claude Code got smarter again in the past month: 1. Most of issues reported by users for PaperFox.ai can be addressed in almost one shot 2. I feel more comfortable to accept the changes and push to main after new and old tests passed (all written by CC) for a feature or bug fix - dev branch has not been used as much as before. But CC still needs human guidance. Just today when I develop the MCP APP for PaperFox.ai: it told me ChatGPT doesn't support MCP APPs yet (wrong), and later hallucinated an incorrect way to render HTML forms for MCP APPs - both fixed when I pointed CC to read the official docs carefully from both OpenAI and Claude. So, coding is largely solved — if you can: - Give correct instructions (right docs, right packages, right context) - Validate the result (automated tests user tests) I'd extend this further: business analytics is largely solved by the same logic - if you know how to ask the right questions and how to validate and interpret the output. The real differentiator — with or without AI — has always been the human's ability to frame the problem and judge the answer. #AIEngineering #ClaudeCode #SoftwareDevelopment #BusinessAnalytics #PaperFox youtube.com/watch?v=We7BZVKb…

I want to thank AI, especially Claude Code. About a year ago, Claude Code was released. I only started trying it in early August 2025 and after two weeks of intensive use, I switched from Cursor and never went back. I've been on the Max plan for 6 months now. In that time, I built PaperFox.ai and launched it in December 2025 to help conference chairs better run their academic conferences. It now serves many conferences all around the world. This feels unreal from a developer standpoint. Software engineering practice, education, and the SaaS industry have been completely disrupted. I want to thank AI for enabling me to do so many things I hoped to do but wasn't capable of doing on my own — it allows me to put my product and design passion into real systems that people actually use - this feels good. #ClaudeCode #aicoding #coding #buildinpublic #AI #SaaS #PaperFox

Technical debt isn't what it used to be. AI is incredibly good at generating technical debt if you don't supervise it. For example, over the last few months, AI happily generated all kinds of variations for empty states and loading spinners in PaperFox.ai —custom styles, inline CSS, slightly different components - it works, but it leaves a mess: inconsistent states across ~165 files. This morning, I guided Claude Code through a massive refactor to standardize everything on simple shadcn/ui components across the entire codebase. I have to say, watching AI systematically clean up this mess (~4,000 lines changed) is extremely fulfilling - I cooked a meal while waiting ;). Normally, this kind of refactoring is tedious, labor-intensive, and doesn’t directly impact functionality so it sits at the bottom of the backlog forever and never gets done. Now with AI, we can actually afford to pay down the technical debt (which, ironically, often created by AI itself). #AI #Refactoring #BuildInPublic #TechnicalDebt #PaperFox

The team here at PaperFox Entertainment would love to congratulate Team Cherry for how amazing Silksong is! #silksong #fanart #hollowknight #silksongfanart #hollowknightfanart #Hornet #Hornetfanart #indiegame #videogames

Hello everyone! We are PaperFox Entertainement and here are some of the object designs you can find in our upcoming game! #gamedev #indiedev #indiegames #indie #videogame #cozygames #animals #wholesomegames #storybook #cardgame #cardbattler #strategy #singleplayer #fantasy

nayami ̶p̶̶e̶̶a̶̶c̶̶h̶ paperfox 👍💫 @NayamiiChan #nayamiart

Мали можливість відправити волонтерські набори на нуль побратимам студії PaperFox Print з Києва, які з самого початку нашої взаємодії допомагали з правками, розуміли ситуацію і робили знижки на волонтерку. Дякуємо всім, хто має змогу звітувати, це дуже нам допомагає!

PAPERFOX made it in … Times Square

In Mumbai at @titlewavesMUM Bandra. Amazon, Flipkart, and Paperfox as well. Reaching more bookstores shortly. @AuthorsInkIndia

‘A Face for You’ - a story of workplace romance and betrayal! Available for pre-order now on Amazon, Paperfox & Rachnaye. <First 100 copies get a specially designed bookmark!> #newbook #readingcommunity #readingfc #Novels #romancebooks #Romance #BookTwitter #bookreading

Do order a copy of The Doctor's Heart Links for purchase- Flipkart dl.flipkart.com/s/una74INNNN Amazon- amzn.to/3RroTnj Paperfox- paperfox.in/product/the-doct… Will soon be Available Worldwide #BookTwitter #Books #WritingCommunity #writerslift #booktwt #booklaunch

My book #DilDuniyaDastoor is available on Amazon, Flipkart & Paperfox. Book your book now 🙏 Flipkart: flipkart.com/dil-duniya-dast… Paperfox: paperfox.in/product/dil-duni… Amazon: amzn.eu/d/bF1ATBJ #NewBook #HindiPoems #Shayari #FirstBook

Presenting a sneak peak into my labour of love…my first book: Need all your love and blessings. Coming Soon ❤️❤️❤️ Pre-release orders can be placed through Amazon. Simply search for the title and place your order. Flipkart and Paperfox will be available soon.

4. Paperfox Coffee Venturing away from campus, this coffee shop is worth the drive. With delicious cold brew and fast wifi, it’s a great place to work on that project you’ve been procrastinating!