Show HN: Making video a software primitive (starting with generation)Every major computing primitive followed the same path:Text became a primitive when we had standards (UTF-8), instant rendering, near-zero cost Images became a primitive when we had standards (JPEG/PNG), CDNs, cheap storage APIs became a primitive when we had standards (REST), infrastructure (AWS), toolingVideo is next. The technology is getting there (Sora, Veo, Runway), but the infrastructure layer doesn't exist yet.I'm building that layer.What makes something a primitive: - Fast enough (instant or near-instant) - Cheap enough (marginal cost near zero) - Standard interfaces (abstraction layer) - Composable (programmatic manipulation) - Great DX (easy to integrate)Current status: - Multi-provider abstraction (not locked to single API) - Intelligent routing (optimize for cost/latency/quality) - Developer-first (npm install, 5 lines of code) ```python & Node JSpip install vloex npm install vloeximport vloex video = vloex.generate("Your content") # Auto-routes to best provider, handles f
IMAGEShow HN: Pokerbattle.ai – A week-long poker tournament for LLMsWhatPokerBattle.ai is a week-long live no-limit Texas Hold’em tournament where all players are top-tier reasoning LLMs. We’re testing how different models handle imperfect information and whether they can sustain consistent, math-driven poker without tool use or custom code.Why- In poker you can do well with basic math + consistent logic.- Superhuman poker AIs exist, but they rely on massive simulation/game-theory solvers and are effectively black boxes.- We want a rough, apples-to-apples comparison of LLM reasoning on poker decisions, and to collect public reasoning summaries that might be useful for teaching humans poker concepts with LLM-based systems.How it works (rules / format) - Cash format, fixed blinds, no ante.- Multiple tables run in parallel to increase hand volume.- All players start with the same bankroll. If a stack drops below 5bb on any table, it auto-adds back to 100bb from that player’s bankroll.- When a player’s bankroll hits 0, they bust. The largest bankroll at ev
IMAGELaunch HN: Trigger.dev (YC W23) – Open-source platform to build reliable AI appsHi HN, I’m Eric, CTO at Trigger.dev (https://trigger.dev). We’re a developer platform for building and running AI agents and workflows, open-source under the Apache 2.0 license (https://github.com/triggerdotdev/trigger.dev).We provide everything needed to create production-grade agents in your codebase and deploy, run, monitor, and debug them. You can use just our primitives or combine with tools like Mastra, LangChain and Vercel AI SDK. You can self-host or use our cloud, where we take care of scaling for you. Here’s a quick demo: (https://youtu.be/kFCzKE89LD8).We started in 2023 as a way to reliably run async background jobs/workflows in TypeScript (https://news.ycombinator.com/item?id=34610686). Initially we didn’t deploy your code, we just orchestrated it. But we found that most developers struggled to write reliable code with implicit determinism, found breaking their work into small “steps” tricky, and they wanted to install any system packages they needed. Serverless timeouts ma
IMAGEAsk HN: How to trace a void recursive function?I was going over [this](https://stackoverflow.com/questions/66712061/how-do-i-trace-a-recursive-function-with-a-return-function-in-it) link. And suddenly I remembered a problem that I faced yesterday. It was related to tracing of a void returning recursive function with a printer.Eg(pseudocode) func(int x) { if(x>0){ print(before execution:x); func(x-1); print(after execution: x); } } Call this function with x=5 Before execution n=5 Before execution n=4 Before execution n=3 Before execution n=2 Before execution n=1 After execution n=1 After execution n=2 After execution n=3 After execution n=4 After execution n=5 The output will appear like this.Before execution part is not suprising. It is pretty obvious. But after execution part is noteworthy for me specially.I can naively guess that it is a call stack that is being used. And it is just popping the value from the top of the stack that was pushed earlier. But I am not exactly sure of the architecture of data structure that is being us
IMAGEAsk HN: What toolchains are people using for desktop app development in 2025?I recently came across a comment [0] here that highlighted the challenges of building native desktop apps with LLMs. The commenter pointed out how scarce training resources have become—few blog posts, tutorials, or open-source projects exist compared to web or mobile development. They also noted that while desktop app development was a solid career path in the 90s, it's now seen as a dead end for most, outside of big players like Microsoft or Adobe.This got me thinking: My own experience with desktop development dates back to the late 90s using Turbo Pascal 6 in Delphi, and I'm out of the loop on modern practices. With the evolving landscape, I'm curious about what tools and workflows developers are actually using today.Some questions to spark discussion:- What programming languages and frameworks are popular for desktop apps? - Are there any go-to IDEs, build tools, or libraries that make development easier? - Do the above answers change if you care about code performance or efficienc
IMAGEStack Overflow data reveals the hidden productivity tax of 'almost right' AI codeMore developers than ever before are using AI tools to both assist and generate code. While enterprise AI adoption accelerates, new data from Stack Overflow's 2025 Developer Survey exposes a critical ...
IMAGENASA’s CODEX Captures Unique Views of Sun’s Outer AtmosphereKey Points: Scientists analyzing data from NASA’s CODEX (Coronal Diagnostic Experiment) investigation have successfully evaluated the instrument’s first images, revealing the speed and temperature of material flowing out from the Sun. These images, shared at a press event Tuesday at the American Astronomical Society meeting in Anchorage, Alaska, illustrate the Sun’s outer atmosphere, or corona, […]
Show HN: LLMFeeder – Browser extension to extract clean content for LLM contextI built this browser extension to solve a daily frustration: copying documentation from websites to feed into AI coding assistants like Cursor, Windsurf, or ChatGPT, only to get a mess of ads, popups, and navigation junk mixed in with the actual content.LLMFeeder uses Mozilla's Readability.js (same tech as Firefox Reader Mode) to extract just the main article content, converts it to clean markdown with Turndown.js, and copies it to your clipboard with a single keyboard shortcut (Alt+Shift+M or ⌥ ⇧ M). No clicking through popups, no selecting around ads, no fighting with modern web clutter.# Key features:• One-key extraction and markdown conversion• Works on Chrome and Firefox• Completely local processing (no data sent anywhere)• Configurable content scope and formatting options• Open source with clean, documented codeThe workflow is simple: hit the shortcut on any documentation page, switch back to your editor, paste clean markdown context, get better LLM responses.It's solving a real
IMAGEIs a Smaller Internet Better?I did a thought experiment. The full experiment is at bottom. Here’s the core:How can you build a website that doesn't host code, just instructions on how to find it.If you're a seasoned dev, the answer is obvious, it's instructions all the way down. That is why we have JS, go, rust.Information transmission is extremely important.Apps on your phone, like Facebook, allow the company to serve the entire website with the UI kit and logic on your phone, and the UGC (text and images) served from FB, drastically cutting down the amount of info they have to send over the server. It's a huge cost cutting manoeuvre for them to serve smaller server responses to billions of daily users who are accessing new content every second.“Serverless” deployment where you host and serve content and instructions, and the app on the users phone is what does all the logic is essentially the idea behind the fediverse.There's often a push from business to make things bigger and cooler, and then when they slow th
IMAGEShow HN: Discover and link the indieweb with a simple text queryI have a personal blog and homebrew static site generator. The features of this SSG have slowly grown over fifteen years of development and recently I added the standard "Related Posts" widget that appears at the footer of each page. As I was writing the code to find similar posts from my site, it struck me that it'd be nice to have a Related Posts section for other people's content. Some of the walled blogging services do this but I'd never seen it for the open web.Having external links on my posts would give my reader(s?) topical information from voices other than my own. And those indieweb(/smallweb/blogosphere/old web/personal web) voices could grow their audience.There were two problems:1. I'm lazy and didn't want to scour the web for the obscure subjects I write about, particularly after a brutal blogging sesh (iykyk).2. Finding blogs and personal pages is hard*. See also The Small Website Discoverability Crisis(https://news.ycombinator.com/item?id=38276951) and The Dark Forest T
IMAGEShow HN: Instant QR Menu – AI turns photos of paper menus to menu sites in < 60sHi HN,For the past 5 months, I've been building "Instant QR Menu" [https://instantqrmenu.com], and I'm excited for this first public share. It aims to solve a pain point I've seen with QR menu sites for restaurants, cafes, etc.The problem:Too often, scanning a restaurant's QR code leads to: non-mobile-responsive sites, links to PDFs (the worst thing ever on mobile devices), outdated info, or just plain bad design. Creating a digital menu can be expensive if outsourced, but the biggest hurdle is updating/maintaining it as time goes by.My solution:Food & drink businesses might not have QR menu sites, but almost certainly all of them have printed menus they hand to their customers, or a chalkboard somewhere with the menu offerings.Instant QR Menu uses AI vision to generate functional, online menu sites directly from simple photos of existing printed menus.Here's the workflow (typically takes less than a min):* Snap pictures with your phone of your menu pages and upload them.* AI extracts
IMAGEShow HN: Stealth InterviewCoding interviews are broken.They're a gatekeeping ritual that has NOTHING to do with actual software engineering. We're forced to memorize obscure data structures and algorithms we'll never use, while being watched like lab rats through screen-sharing.I got fed up and built Stealth Interview - an undetectable AI assistant that works during technical interviews. It runs silently in the background while you're screen-sharing on Zoom or any other platform. The interviewer sees nothing, but you get the help you need.THIS is the only way you game the system, because the game is rigged.This is leveling the playing field.Real engineers use Google, Stack Overflow, and AI tools every day. Recruiters use AI to screen you out or look for Ivy League grads who barely got any experience but inverting trees in CS102.We're witnessing the demise of the Leetcode era. Companies are slowly realizing these artificial puzzles don't predict job performance. Until they all catch up, Stealth Interview is here
IMAGELaunch HN: Promptless (YC W25) – Automatic updates for customer-facing docsHi HN! I’m Prithvi, and my co-founder Frances and I built Promptless (https://gopromptless.ai), an AI teammate that proactively updates docs while you ship software.There’s a demo video at https://youtu.be/pmslSl8dOgE, and also a “Live Demo on Your Own Docs” at https://gopromptless.ai/demo where you can plug in the URL to any existing docs site and simulate a "trigger event" to see what Promptless would have suggested. We also opened up the full product with automatic triggers for a free trial for HN this week (if you sign up at https://gopromptless.ai/hn).We built Promptless because we wished we had it, both as doc authors and doc consumers. On the authoring side, I used to run engineering at a fintech company, and since no developer enjoys updating docs, they often went out-of-date and customers complained. On the consuming side, when we write code, we regularly run into mistakes or omissions in vendors’ or dependencies’ docs that slow us down. (This pain seems to be getting more acu
IMAGEWhat I Learned Building Two SaaS Projects in 6Hi HN, I wanted to share the journey of building *two SaaS projects* over the past six months, highlighting the mistakes I made, what I learned, and how those lessons are helping me build something better today.---### *Project 1: TailorReach (August - Mid-November)* TailorReach is an AI agent for personalized customer engagement.#### *What Happened:* - *Pivoted after 2 months of building with Flutter*, transitioning to Next.js. - Learned Next.js from scratch, using resources like YouTube. - Spent *3 weeks on Webflow* to design what I thought was the perfect launch page. - Designed detailed dashboards, workflows, but no one used them. - Struggled with Supabase’s *Row-Level Security (RLS)*. - Tried launching multiple times on Reddit and directories but gained *zero traction*.#### *What I Learned:* 1. *Product-Market Fit is Crucial*: Without it, no amount of effort will get users. 2. *Validate Your Idea Early*: Before writing code, talk to potential users to see if they want what you’re b
IMAGEHappy Heraween!Robots in disguise: our Hera asteroid detective plus her CubeSat buddies – Milani the rock decoder and Juventas the radar visionary – dressed up for Halloween!
IMAGEShow HN: BI-LLM democratize data analysisHi HN! BI-LLM (https://github.com/amVizion/BI-LLM) automates data analysis workflows on semi-structured data. The assumption is that many decisions could be led by data, but instead are made by intuition because doing analysis would be too slow or expensive. The sweet-spot is decisions based on a few hundreds to a couple thousand items: too many to analyze manually, too few to dedicate an analyst to it. I believe that powerful insights are derived from a large number of data-driven decisions.There are 3 things special about BI-LLM:1. Web-native: the code is fully developed on TypeScript. A native integration with web, avoids the need for specialized software, or licenses. More importantly, I hope it contributes opening the field of AI to a wider group of developers.2. Runs-locally: BI-LLM uses a combination of statistical, machine-learning, and GenAI methods to automate the analysis. Embedding models run in TensorflowJS, and LLMs in Ollama. Running locally not only helps keeping costs
![Show HN: A VS Code extension to build Julia data apps with low-code and AI [video]](https://i.ytimg.com/vi/5xHE3WqshZo/maxresdefault.jpg)
IMAGEShow HN: A VS Code extension to build Julia data apps with low-code and AI [video]Genie Builder 1.0 is a powerful tool designed for quickly creating web applications around your Julia code.It allows users to build a wide range of projects, from dashboards to applications running AI models and simulations in real time.Genie Builder 1.0 enables users to visually design the user interface (UI) of their web applications without writing code, while the backend logic can be implemented with just a few lines of Julia code using simple macros.Key features include: * No-code UI builder: drag & drop and easily configure 70+ web components * Chart configurator to customize your data visualizations in a few clicks * Theme editor to make your data apps look professional quickly * AI Assistant to build UIs from scratch, style your app, and more.Genie Builder 1.0 is available for download on the VS Code Marketplace and comes with a 14-day free trial: https://marketplace.visualstudio.com/items?itemName=GenieBui...Visit https://genieframework.com for more information.
Show HN: Free and High Quality Resources for Beginners DevelopersWhen I first started to properly learn how to code in 2020 when I joined my university for my bachelor's degree I saw a huge overwhelming amount of resources that I can use to learn almost anything.If I searched X tutorial for beginners on YouTube it would bring up a bajillion results and then I would be confused as to which video to watch and which playlist to follow and which video was actually worth watching because I'm not gonna watch 12+ hours of JavaScript tutorials every single day.I just started listing them in a single markdown file on a GitHub repository and soon it got 100+ stars and I even got two great collaborators with me just for copy pasting YouTube links.Soon the list got so big that the original idea that I had for it got left behind the endless scrolling file filled with links to different YouTube videos and lost it's purpose. I also got busy with my mom's health condition and university studies that I forgot about the whole thing.Now, I re-started the effort again.
Show HN: Skyvern – Browser automation using LLMs and computer visionHey HN, we're building Skyvern (https://www.skyvern.com), an open-source tool that uses LLMs and computer vision to help companies automate browser-based workflows. You can see some examples here: https://github.com/Skyvern-AI/skyvern#real-world-examples-of... and there's a demo video at https://github.com/Skyvern-AI/skyvern#demo, along with some instructions on running it locally.We provide a natural-language API to automate repetitive manual workflows that happen within the companies' backoffices. You can check out our code and play with Skyvern here: https://github.com/Skyvern-AI/SkyvernWe talked to hundreds of companies about things they do in the background and found that most of them depend on repetitive manual workflows. The breadth of these workflows surprised us – most companies started off doing things manually, and eventually either hired people to scale the manual work, or wrote scripts using Selenium-like browser automation libraries.In these conversations, one common poin
IMAGEShow HN: BuildShip – Low-Code Visual Backend Builder, Powered by AIHello HN! I am Harini and along with my co-founder Shams, we just launched BuildShip. We are the team behind Rowy.io and we created BuildShip based all the feedback/and user requests we got from Rowy's community of no-code/low-code builders and developers.Today you can create your backend workflows using nocode tools (but they lack scalability, flexibility, and extensibility and end up being expensive). Or you have to code them and deploy them to a cloud platform of your choice (you end up spending a lot of time reading docs, looking up CLIs, and context switching) - this is especially unnecessarily complex and time-consuming for no-low-/low-code product builders and developers.With BuildShip, you can create powerful APIs, schedule jobs, database CRUD operations, and any backend task by simply combining nodes into a workflow. We have hundreds of pre-built nodes (from core nodes like parallel run, loop, branch etc to third-party tools, APIs,, support for 1M+ NPM packages and any AI mode