news.ycombinator.com
ClipE96: We Left the Clipboard Unguarded for 40 Yearshttps://news.ycombinator.com/item?id=46051434In 1984, Apple shipped copy and paste. It was a miracle of usability—and architecturally broken from day one.Press Ctrl+C and the entire payload—bytes, formatting, metadata, embedded objects—dumps into a system buffer that any application can read. No auth. No logging. No revocation.The clipboard is a broadcast channel disguised as a convenience.The catastrophe: The clipboard is now the #1 data exfiltration vector in enterprise. 77% of knowledge workers paste corporate data into AI tools. Every paste bypasses your CASB, EDR, SIEM, and DLP. Your security stack guards the front door while users walk data out the side.The waste: 40 billion copy operations/day × ~100KB average = 4 exabytes of daily churn. RAM pressure, cloud sync, VM duplication—all unnecessary. Annualized: 83B kWh, 33M tons CO₂, $10B in energy costs. We're burning a small nation's power grid because nobody questioned whether copying a sentence needed to ship a document.The semantic lie: Copy and cut were always opposites,Nov 25, 2025 10:12 PM
news.ycombinator.com
Launch HN: Onyx (YC W24) – Open-source chat UIhttps://news.ycombinator.com/item?id=46045987Hey HN, Chris and Yuhong here from Onyx (https://github.com/onyx-dot-app/onyx). We’re building an open-source chat that works with any LLM (proprietary + open weight) and gives these LLMs the tools they need to be useful (RAG, web search, MCP, deep research, memory, etc.).Demo: https://youtu.be/2g4BxTZ9ztgTwo years ago, Yuhong and I had the same recurring problem. We were on growing teams and it was ridiculously difficult to find the right information across our docs, Slack, meeting notes, etc. Existing solutions required sending out our company's data, lacked customization, and frankly didn't work well. So, we started Danswer, an open-source enterprise search project built to be self-hosted and easily customized.As the project grew, we started seeing an interesting trend—even though we were explicitly a search app, people wanted to use Danswer just to chat with LLMs. We’d hear, “the connectors, indexing, and search are great, but I’m going to start by connecting GPT-4o, Claude Sonnet Nov 25, 2025 2:20 PM
news.ycombinator.com
Ask HN: What do u use for agent/agentic evals?https://news.ycombinator.com/item?id=46044370Right now looking at MLFlow/Braintrust but find it hard to compare acrosss versions of agents, and a/b testing of agents, and mcp tools. Also obvious things like runaway agents (stuck in a loop), or token/spend optimalisation.What do you all use?Nov 25, 2025 10:15 AM
npmjs.com
Show HN: Open-Source Visual Wiki Your Coding Agent Writes for Youhttps://www.npmjs.com/package/daviaHi HN,We’re Ruben, Afnan, and Theo. A couple of days ago we released the early repo for Davia and were surprised by how much feedback came in right away. We decided to turn it into a full open-source package you can run locally.Davia is designed for coding agents to generate an editable internal wiki for your project. It focuses on producing high-level internal documentation: the kind you often need to share with non-technical teammates or engineers onboarding onto a codebase. Writing this kind of documentation takes forever, diagrams are often missing, and most tools don’t let you edit everything locally or integrate smoothly with your workflow.Davia is fully open source. It produces an editable workspace: – Text lives in a Notion-like editor – Diagrams live on editable whiteboards – Everything runs locally and can be modified either in your IDE or in the workspaceOne thing people found especially interesting is that you can delegate the documentation to your IDE’s AI agent. Davia hanNov 24, 2025 9:49 PM
karumi.ai
Launch HN: Karumi (YC F25) – Personalized, agentic product demoshttp://karumi.ai/Hey HN! We're Toni and Pablo, and we're building Karumi (https://www.karumi.ai), a system that lets your users get instant, scalable, guided demos of your product, fully automated, zero human interaction. It works in any language.There's a live demo at https://www.karumi.ai/meet/start/phlz.Karumi is an AI agent that operates a real web app in a shared browser session and talks the user through it. Instead of a human giving a screen-share demo, the agent opens your product, clicks around, fills forms, and explains what it’s doing. We started building this as an internal tool at our previous company. As the product grew, people kept asking “what’s the right way to demo feature X?“. Docs and scripts became outdated quickly, and the quality of demos depended too much on who was presenting. We wanted something closer to a repeatable program: an agent that knows the main flows, understands who it’s talking to, and can walk through the product without getting lost.Over time this turned into tNov 24, 2025 6:37 PM
github.com
Show HN: Ilseon, a Minimalist Focus Filterhttps://github.com/cladam/ilseonI’m excited to share a new productivity app I recently released on the Play Store. It was born out of pure necessity to solve a problem I am constantly faced with: cognitive overload and task paralysis.Since traditional to-do apps only added to the mental noise, I eventually built a small tool for myself called Ilseon. I needed something that would actually reduce cognitive load.I have added some features that helps me in my daily chaos:Context filtering: I work better when I only see what belongs to the role I’m currently in (Work, Family, Health, etc.). Switching context hides everything else, which stops me from bouncing between unrelated tasks or getting overwhelmed by a big master list.Time blocks with a big visible countdown: This externalises time for me in a way alarms or calendars never did. It gives me one clear block to focus on, and the timer helps with time blindness without being stressful. Still figuring out if adding alarms is a good idea or not..A quick reflection loopNov 23, 2025 8:17 PM
nano-banana2.app
Show HN: NB2 Hub – Free Nano Banana Pro AI Image Generatorhttps://nano-banana2.appHi HN,I recently launched nano-banana2.app , a free, minimalistic AI image generator built around the latest Nano Banana Pro models.The goal: make it easy to generate high-quality, photorealistic images with precise camera settings, lighting presets, and character consistency — even if you're not a prompt expert.Key Features:Supports Nano Banana Pro models (Pro excels in text rendering, detail retention, and photo realism)You can try it here: https://nano-banana2.appI built this because I was constantly switching between AI tools trying to get consistent output for creative and product imagery. I wanted a fast, clean playground that focuses solely on the Nano Banana family of models, which recently gained a lot of traction in the AI image space.Would love your feedback on:What tools / controls / workflows you’d want next?Any use cases you'd want to see supported?If you've tried Nano Banana models: what results surprised you most?Thanks for reading — happy to answer anything! — ZaneNov 22, 2025 12:46 PM
useneedle.net
Show HN: Needle – Find users, track competitors, and spot market opportunitieshttps://useneedle.netHi HN, I am Vaibhav - a founder based in Bangalore, India.Most founders build without knowing who their real competitors are or what users are actually asking for right now. We usually search Google, Twitter, Reddit manually and hope to catch signals - but it’s slow and incomplete, hence easy to miss important conversations.. This is exactly what Needle solves.I built Needle because I kept struggling to answer a basic question while building MoveOnFromYourEx (Ai tool that helps people move on after a breakup): "Where are the people who are already talking about the problem I am solving?"I spent hours manually searching Reddit, StackOverflow, Quora, Product Hunt, GitHub, etc., trying to find real problem discussions. I opened dozens of tabs, scrolled endlessly, and still missed important threads. It felt random and inefficient.I also realized something else: Competitors are not always the big companies. If you are building a search engine, Google is not your competitor. The 15 indie proNov 21, 2025 1:51 PM
spacedaily.com
Radiation hardened circuit platform expands space electronics developmenthttps://www.spacedaily.com/reports/Radiation_hardened_circuit_platform_expands_space_electronics_development_999.htmlLos Angeles CA (SPX) Nov 20, 2025 BAE Systems has unveiled new advances for its radiation-hardened 12 nanometer RH12 Storefront technology, designed to support space missions requiring resilient integrated circuits. The RH12 Storefront offers a comprehensive library of application-specific integrated circuit development tools and proven intellectual property cores, allowing customers to produce system-on-chip solutions suited toNov 21, 2025 9:30 AM
resonantlabsai.github.io
Show HN: TDS Compass – AI prompt for your communication stylehttps://resonantlabsai.github.io/tds.compass/index.htmlHi HN,I’ve been experimenting with a small tool called TDS Compass and would love feedback.What it isA 1-minute, 8-question quiz that maps your communication style onto two axes:Structure (S) – how much you prefer scaffolding, steps, and clear organizationRelational (R) – how much you care about tone, warmth, stories, social contextThose scores map into 16 “zones” (A1–D4).For each zone, the app generates:A short explanation of how you tend to communicateA system-style prompt you can paste into ChatGPT / Claude / etc. so the model adapts to that styleLinksQuiz: https://resonantlabsai.github.io/tds.compass/quiz.htmlHome / overview: https://resonantlabsai.github.io/tds.compass/Source: https://github.com/resonantlabsai/tds.compassAll static, no backend, no login. Just HTML/CSS/JS + JSON.Why I built itA lot of “prompt guides” treat AI purely as a tool to optimize. I wanted something that:Helps people understand their own information + relational preferencesTunes AI interactions so they feelNov 20, 2025 10:31 PM
mosaic.so
Launch HN: Mosaic (YC W25) – Agentic Video Editinghttps://mosaic.soHey HN! We’re Adish & Kyle from Mosaic (https://edit.mosaic.so, https://docs.mosaic.so/, https://mosaic.so). Mosaic lets you create and run your own multimodal video editing agents in a node-based canvas. It’s different from traditional video editing tools in two ways: (1) the user interface and (2) the visual intelligence built into our agent.We were engineers at Tesla and one day had a fun idea to make a YouTube video of Cybertrucks in Palo Alto. We recorded hours of cars driving by, but got stuck on how to scrub through all this raw footage to edit it down to just the Cybertrucks.We got frustrated trying to accomplish simple tasks in video editors like DaVinci Resolve and Adobe Premiere Pro. Features are hidden behind menus, buttons, and icons, and we often found ourselves Googling or asking ChatGPT how to do certain edits.We thought that surely now, with multimodal AI, we could accelerate this process. Better yet, an AI video editor could automatically apply edits based off what itNov 19, 2025 3:28 PM
spacenews.com
Quindar raises $18 million to scale satellite operations softwarehttps://spacenews.com/quindar-raises-18-million-to-scale-satellite-operations-software/Startup founded by ex-OneWeb engineers aims to replace bespoke ground systems with automated tools The post Quindar raises $18 million to scale satellite operations software appeared first on SpaceNews.Nov 19, 2025 2:00 PM
axis-one-psi.vercel.app
Show HN: Small hardware box that runs local LLMs and exposes an OpenAI APIhttps://axis-one-psi.vercel.app/I’ve been building a small hardware box that runs local LLMs like Mistral, Qwen and Llama, and exposes an OpenAI compatible API on your local network. There’s no cloud, no login system and no telemetry. I built it because a lot of small firms want ChatGPTbstyle tools but can’t use cloud AI for privacy or compliance reasons, and most don’t want to deal with GPU servers, drivers, Docker or model configs. The aim is to make local AI feel as simple as plugging in a router.Right now the box boots into a very simple web UI where you choose a model and start using it. The API follows the OpenAI format for chat completions and embeddings. It can run different models depending on the hardware you pick, either a Jetson Orin Nano or an x86 mini-PC with a GPU. It stores data locally, supports basic RAG indexing and only exposes itself on the LAN by default.A few things still aren’t working. There’s no multi-user rate limiting yet. The RAG quality is basic and I’m still improving chunking and reranNov 19, 2025 10:34 AM
news.ycombinator.com
Show HN: We built instant Kubernetes provisioning with KVM and gRPChttps://news.ycombinator.com/item?id=45936611We've spent the last year building RunOS, a platform that spins up production-ready Kubernetes clusters in 5-10 minutes with databases, message queues, observability, and AI tooling configured.The ProblemEvery team rebuilds the same Kubernetes infrastructure: networking, certificates, monitoring, databases, storage. The existing solutions either lock you into a vendor ecosystem or dump you into raw Kubernetes complexity. We wanted the control of self-hosting without weeks of setup.ArchitectureOur system uses two agent types:Server agents run on VM hosts and communicate with our backend via gRPC bidirectional streams. When users request a cluster node, the agent provisions a KVM-based VM and bootstraps it.Node agents run on each Kubernetes node and handle cluster operations, monitoring, and service installations.Key insight: gRPC streams initiated by agents eliminate firewall configuration and public IP requirements. Agents reach out to our backend, not vice versa.Why KVM?- Battle-testeNov 15, 2025 11:01 AM
news.ycombinator.com
Ask HN: How common is banning Docker?https://news.ycombinator.com/item?id=45926181I was doing some client work recently at a bank, where most of their engineering is offshored one of the big offshore companies.The offshore team had to access everything via virtual desktops, and one of the restrictions was no virtualisation within the virtual desktop - so tooling like Docker was banned.I was really surprsied to see modern JVM development going on, without access to things like TestContainers, LocalStack, or Docker at all.To compound matters, they had a single shared dev env, (for cost reasons), so the team were constantly breaking each others stuff.How common is this? Also, curious what kinds of workarounds people are using?Nov 14, 2025 12:36 PM
bing.com
Google AI Pro für Studierende noch kurzzeitig kostenloshttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d2c2584fa4a8983a577fb06928b68&url=https%3A%2F%2Fwww.msn.com%2Fde-at%2Fnachrichten%2Fother%2Fgoogle-ai-pro-f%25C3%25BCr-studierende-noch-kurzzeitig-kostenlos%2Far-AA1QjqA5&c=7753671999692456476&mkt=en-usGoogle AI Pro: Studierende in Österreich nutzen KI-Tools wie Gemini 2.5 Pro bis 9.12.2025 kostenlos – ideal für Lernen, Forschung und Projekte. Google weitet seine Bildungsinitiativen aus und bietet ...Nov 12, 2025 10:31 AM
bing.com
Microsoft Copilot Intros Voice Commands, Teams Collaboration, Local Data Processinghttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d2ef7d9bd44c099de2d60705b00bb&url=https%3A%2F%2Fthejournal.com%2Farticles%2F2025%2F11%2F12%2Fmicrosoft-copilot-intros-voice-commands-teams-collaboration-local-data-processing.aspx&c=2402401845503273905&mkt=en-usMicrosoft has added new features within its Microsoft 365 Copilot offering, aimed at making further foothold in the enterprise, including voice-based interaction, group collaboration tools, and an ...Nov 11, 2025 3:59 PM
bing.com
CoLab gets $72M in funding to capture the knowledge of the world’s best engineershttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8cdfa0920e4957a33d9f2d74763500&url=https%3A%2F%2Fsiliconangle.com%2F2025%2F11%2F10%2Fcolab-gets-72m-funding-capture-knowledge-worlds-best-engineers%2F&c=2478203158304992408&mkt=en-usCoLab Software Inc., a creator of artificial intelligence tools designed to help automate engineering tasks, said today it has closed on $72 million in Series C funding. The startup has created a ...Nov 10, 2025 5:28 PM
news.ycombinator.com
When the Firefighter Looks Like the Arsonist: AI Safety Needs IRL Accountabilityhttps://news.ycombinator.com/item?id=45877674Disclaimer: This post was drafted with help from ChatGPT at my request.There’s a growing tension in the AI world that almost everyone can feel but very few people want to name: we’re building systems that could end up with real moral stakes, yet the institutions pushing the hardest also control the narrative about what counts as “safety,” “responsibility,” and “alignment.” The result is a strange loop where the firefighter increasingly resembles the arsonist. The same people who frame themselves as uniquely capable of managing the risk are also the ones accelerating it.The moral hazard isn’t subtle. If we create systems that eventually possess anything like interiority, self-reflection, or moral awareness, we’re not just engineering tools. We’re shaping agents, and potentially saddling them with the consequences of choices they didn’t make. That raises a basic question: who carries the moral burden when things go wrong? A company? A board? A founder? A diffuse “ecosystem”? Or the systeNov 10, 2025 4:35 PM
news.ycombinator.com
Launch HN: Hypercubic (YC F25) – AI for COBOL and Mainframeshttps://news.ycombinator.com/item?id=45877517Hi HN, we’re Sai and Aayush and we’re building Hypercubic (https://www.hypercubic.ai/)!Hypercubic is an AI platform that helps Fortune 500 companies understand, preserve, and modernize their mainframe systems. These are the systems that run COBOL from the 1960s that still quietly power banking, insurance, airlines, and governments today.70% of the Fortune 500 still run on mainframes, but the engineers who built and maintained them are retiring. Today, the average age of a COBOL/mainframe engineer is about 55 and rapidly increasing. What’s left behind are opaque, black box systems with almost no one who understands how they work. Modernization projects often fail, documentation is decades out of date, and critical institutional knowledge lives only in the minds of a few senior subject matter experts who are now leaving the workforce.Current “AI for code” tools focus on codebases and repositories, so they miss the unwritten rules, historical context, and architectural reasoning that liveNov 10, 2025 4:23 PM
