bing.com
'Project Hail Mary's Directors Want To Adapt This Impressive Sci-Fi Book Next After Their 95% RT Hithttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8cece2c3b645b1946679ddd8063fd4&url=https%3A%2F%2Fcollider.com%2Fproject-hail-mary-phil-lord-chris-miller-artemis-adaptation-andy-weir%2F&c=17748582200535083972&mkt=en-usJulio is a Senior Author for Collider. He studied History and International Relations at university, but found his calling in cultural journalism. When he isn't writing, Julio also teaches English at ...Apr 5, 2026 4:08 AM
news.ycombinator.com
it's not Ai if the LLM is not in controlhttps://news.ycombinator.com/item?id=47642344I always thought that the frontend of "Ai" is awful, but now I know it for sure:OAI5.1+ is good, but chatgpt sucks, it doesn't have gmail integration and barely able to do anything but basic retrieval from the integrations it actually has.Opus is amazing, but claude web is mediocre at best. It has a very limited set of integrations even after 2 years, some don't even work (clay), and it uses way too many tokens to do basic stuff.XAi is ok for social queries but grok is very bad. Its memory is basic, and the grok teams takes ship features 18 months later.in 2024, i thought the problem is that all of this is new. "they just need a little more time" I told myself, but the truth is that the scaffolding is truly rubbish. Other the claude (which is barely good), these products are not what we were promised in this cycle. If we allowed any Ai, even the most basic models to build the frontend for chatgpt or grok, it would have done a better job! At the very least it would have shipped more feaApr 4, 2026 7:19 PM
bing.com
Trent Perry announces return to UCLA, bolstering backcourthttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d4a918a5d4aa9bf2dc4eed196bec7&url=https%3A%2F%2Fwww.aol.com%2Fsports%2Ftrent-perry-announces-return-ucla-021743534.html&c=13118344956345421173&mkt=en-usTo those who say Mick Cronin is a coach few want to play for, that he’s too demanding and volatile, Trent Perry was just the latest to provide a retort in all capital letters. “I’M BACKKKK,” the UCLA ...Apr 4, 2026 7:17 PM
github.com
Show HN: Mason – A container for running a team of agents using a conciergehttps://github.com/Mason-Teams/mason-teamsWhile working on a few projects with claude code I found myself building a bunch of scaffolding to organize and build my agent workflow. Trying to figure out how to orchestrate my agents and work with them in a more natural way was more interesting to me. This container has all of the basic building blocks of my workflow. I use both the terminal and chat server depending on what I’m trying to do. So do what feels comfortable to you. I suggest having the terminal window open and watching the agents work is a good idea to get a feel for things. Take a look at the README and the other docs. I’ll be checking this thread and will respond when I can. Thanks for reading this far!Apr 4, 2026 3:52 PM
news.ycombinator.com
My 11-step GraphRAG pipeline, what worked, and what's still brokenhttps://news.ycombinator.com/item?id=47639059While building a financial assistant for an SF start-up, we learned that AI frameworks add complexity without value. When I started building a personal assistant with GraphRAG, I carried that lesson but still tried LangChain's MongoDBGraphStore. It gave me a working knowledge graph in 10 minutes.Then I looked at the data. I had 17 node types and 34 relationship types from just 5 documents, including three versions of "part of". GraphRAG is a data modeling problem, not a retrieval problem.The attached diagram shows the full 11-step pipeline I ended up with. Here is a walkthrough of what you can learn from each step.So basically, in steps 1 and 2 of the data pipeline, raw sources go through an Extract, Transform, Load (ETL) process. They land as documents in a MongoDB data warehouse. Each document stores the source type, URI, content, and metadata.Then in step 3, we clean the documents and split them into token-bounded chunks. We started with 512 tokens with a 64-token overlap. Still, weApr 4, 2026 1:50 PM
github.com
Show HN: Batty – Run a team of AI coding agents in tmux with test gatinghttps://github.com/battysh/battyHi HN, I'm the author.I use Claude Code and Codex daily. Running one agent on a task works great. Running three or four in parallel on the same repo? They step on each other's files, nobody checks if the code compiles, and you spend more time coordinating than coding.Batty is the supervisor layer I built to fix this. You define a team in YAML — an architect that plans work, a manager that dispatches it, engineers that execute. Batty launches each role in its own tmux pane, isolates engineer work in git worktrees, routes messages between roles, and gates task completion on passing tests.The interesting part is what it's not: it's not an agent framework, and it doesn't embed any model. It orchestrates existing agent CLIs (Claude Code, Codex, Aider) using tmux as the runtime and git worktrees for isolation. Config is YAML, the kanban board is Markdown (powered by a bundled kanban-md tool), inboxes are Maildir, logs are JSONL. You can `git diff` your entire team state.Built in Rust, publisApr 4, 2026 1:02 PM
bing.com
[Photo] Bisly delivers strong pitchinghttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d95a657eb4b7080899cbae36d8b09&url=https%3A%2F%2Fbiz.chosun.com%2Fen%2Fen-photo-spo%2F2026%2F04%2F04%2FDLUBPXOFKZCDRP4OTIBTZR6K7A%2F&c=6547656343909758518&mkt=en-usOn the 4th at Sajik Baseball Stadium in Busan, the 2026 Shinhan SOL KBO League game between the Lotte Giants and the SSG Landers was held. For the home team Lotte, Beazley started, and for the ...Apr 4, 2026 10:56 AM
bing.com
[Photo] Bisly delivers strong pitchinghttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d873d7d874954aeee218096cf21a9&url=https%3A%2F%2Fbiz.chosun.com%2Fen%2Fen-photo-spo%2F2026%2F04%2F04%2FDLUBPXOFKZCDRP4OTIBTZR6K7A%2F&c=6547656343909758518&mkt=en-usOn the 4th at Sajik Baseball Stadium in Busan, the 2026 Shinhan SOL KBO League game between the Lotte Giants and the SSG Landers was held. For the home team Lotte, Beazley started, and for the ...Apr 4, 2026 10:56 AM
bing.com
Finding a reliable Qibla app for daily prayer—wherever you arehttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8cd0a6472b4784b95428b373a801c4&url=https%3A%2F%2Fwww.dhakatribune.com%2Ffeature%2Ftech%2F406927%2Ffinding-a-reliable-qibla-app-for-daily&c=10645701448462492131&mkt=en-usDaily prayer often happens in changing locations—home, office, hotels, or while traveling. The key challenge is ensuring Qibla guidance stays accurate and reliable anywhere. Travel naturally ...Apr 4, 2026 9:20 AM
news.ycombinator.com
Tell HN: YouTube history setting privacy implicationshttps://news.ycombinator.com/item?id=47636933As user viewer (not channel owner) I have setup YouTube history to be only for max 3 months. Then auto-delete. Today I checked that this setting applies only to users' Video Watch History.One has to manually go https://myactivity.google.com/page?page=youtube_comment_likes to delete Your Likes and Dislikes on YouTube anddelete comments at https://myactivity.google.com/page?hl=en&utm_medium=web&utm_source=youtube&page=youtube_comments for deleting Youtube Comments.Apr 4, 2026 8:01 AM
github.com
Show HN: Travel Hacking Toolkit – Points search and trip planning with AIhttps://github.com/borski/travel-hacking-toolkitI use points and miles for most of my travel. Every booking comes down to the same decision: use points or pay cash? To answer that, you need award availability across multiple programs, cash prices, your current balances, transfer partner ratios, and the math to compare them. I got tired of doing it manually across a dozen tabs.This toolkit teaches Claude Code and OpenCode how to do it. 7 skills (markdown files with API docs and curl examples) and 6 MCP servers (real-time tools the AI calls directly).It searches award flights across 25+ mileage programs (Seats.aero), compares cash prices (Google Flights, Skiplagged, Kiwi.com, Duffel), pulls your loyalty balances (AwardWallet), searches hotels (Trivago, LiteAPI, Airbnb, Booking.com), finds ferry routes across 33 countries, and looks up weird hidden gems near your destination (Atlas Obscura).Reference data is included: transfer partner ratios for Chase UR, Amex MR, Bilt, Capital One, and Citi TY. Point valuations sourced from TPG, UpgraApr 4, 2026 2:26 AM
bing.com
UCLA’s Trent Perry staying for his junior seasonhttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d4a918a5d4aa9bf2dc4eed196bec7&url=https%3A%2F%2Fwww.ocregister.com%2F2026%2F04%2F04%2Fuclas-trent-perry-staying-for-his-junior-season%2F&c=17570253479202739498&mkt=en-usLOS ANGELES — Mick Cronin’s pitch to keep sophomore Trent Perry at UCLA started well before the offseason. After the UCLA men’s basketball team beat Northwestern on Jan. 24, the Bruins’ head coach ...Apr 4, 2026 1:57 AM
news.ycombinator.com
Show HN: EU Compliance SaaS for Sale ($4K Each) – CBAM, AI Act, Public Tendershttps://news.ycombinator.com/item?id=47632626I built 4 SaaS products targeting mandatory EU regulations. Each for $4,000.1. CBAM OS (cbam-os.com) - EU Carbon Border Tax compliance for importers 2. AIA Proof (aiaproof.com) - AI Act compliance & AI detection 3. AO France (ao-france.fr) - AI-powered public tender responses (FR) 4. AO Copilot (ao-copilot.fr) - BTP tender analysis, 20 modules, 1900+ testsAll mandatory regulations = guaranteed market. Modern stack (Next.js, Supabase). Production-ready.Contact: abbas.billel1985@gmail.comApr 3, 2026 9:36 PM
news.ycombinator.com
ARCHE3-7B – Sparse Moe with SmartRouter and Foundation Curriculum Traininghttps://news.ycombinator.com/item?id=47628238This is my first post on HN — a bit nervous, but excited to share what I've been building.I’ve been working on a 7B sparse Mixture-of-Experts prototype that can actually run on consumer hardware. For example, on a Colab T4 it uses around 5 GB RAM and 5 GB VRAM during training, and roughly 3.5–5 GB for inference.A couple of things I spent a lot of time on:Routing (SmartRouter) I tried to tackle routing collapse in a practical way. Instead of letting all tokens dump into a few "favorite" experts, I combined a few things: load balancing loss, an entropy bonus to keep the distribution flat, jitter noise during training, and a learnable temperature. It works surprisingly well at keeping a good portion of experts active. I’ve open-sourced the router code (hive_router.py) if anyone wants to look at the math or grab it for their project.Foundation Curriculum Training (FCT) Before standard pretraining, I run the model through structured reasoning patterns — currently 290 of them across 14 cogniApr 3, 2026 3:54 PM
github.com
Show HN: Nova v1.5.0 Gold Master – A pure C++17 assistant with zero dependencieshttps://github.com/94BILLY/NOVA/releasesI posted an early version of NOVA about a month ago. Since then, I’ve hit the v1.5.0 Gold Master milestone. I’ve spent the last few weeks refining the low-level efficiency and stripping out every possible byte of bloat to keep it as a 'clean room' Win32 implementation. It’s been a challenge to maintain high-DPI support while keeping that Classic aesthetic, but it's finally ready.Apr 3, 2026 3:50 PM
spacescout.info
Artemis II Sets Course for Moonhttps://www.spacescout.info/2026/04/artemis-ii-sets-course-for-moon/After a picture-perfect launch Wednesday placed the crew of Artemis II into a highly elliptical Earth orbit, astronauts Reid Wiseman, Victor Glover, Jeremy Hansen, and Christina Koch are now officially on their way towards the Moon, the first humans to make such a journey since 1972.Apr 3, 2026 3:22 PM
spacenews.com
Europe’s strategic autonomy in space will define its role in the ‘second space age’https://spacenews.com/europes-strategic-autonomy-in-space-will-define-its-role-in-the-second-space-age/Europe’s future in space really boils down to one question: can it stay ahead without relying on technology made somewhere else? As we step into what experts call the “second space age,” strategic autonomy is suddenly front and center for the European Union. And it is increasingly essential that Europe can get to, control and […] The post Europe’s strategic autonomy in space will define its role in the ‘second space age’ appeared first on SpaceNews.Apr 3, 2026 11:00 AM
bing.com
These Are the Top 10 Best Games Ever Made (According to You)http://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8d0bb094ba4aa6b2e72472fdd32eaa&url=https%3A%2F%2Fgamerant.com%2Ftop-10-best-video-games-all-time%2F&c=8995010891483235934&mkt=en-usDalton Cooper is the Managing Editor of GameRant. Dalton has been writing about video games professionally since 2011 and joined the GameRant team in 2015. Having written thousands of game reviews and ...Apr 3, 2026 9:00 AM
bing.com
Roof Box or Hitch Carrier? The MPG Difference Might Surprise Youhttp://www.bing.com/news/apiclick.aspx?ref=FexRss&aid=&tid=6a8cca7e7ca04dd390a48be91eba5f65&url=https%3A%2F%2Fwww.caranddriver.com%2Fcar-accessories%2Fa70910309%2Froof-box-vs-hitch-cargo-carriers-fuel-economy%2F&c=17858090157698061107&mkt=en-usIf you're packing up the family for a long road trip, it doesn't take long to run out of space inside a vehicle. At a certain point, extra storage becomes a necessity. After all, no one wants to spend ...Apr 3, 2026 7:50 AM
animateoldphotos.org
Show HN: I thought merging two photos with AI would be a weekend project. Nopehttps://animateoldphotos.org/add-loved-one-to-photoAt a friend's gathering, someone mentioned wanting to add her late father into a family photo. I figured this would be trivial — modern image models are powerful, just send two photos and ask the AI to merge them. She said she'd tried, but the results were weird and unstable, and she didn't know how to describe what she wanted.I went home and tried it myself. With a well-written prompt and two good photos, it works. But real-world use cases aren't two good photos — it's a modern family photo plus a damaged old portrait, or two old photos from different decades. That's when things fall apart.I looked at existing tools. Most showcase merges between clean, well-lit, modern photos. Nobody was solving the hard version: mismatched eras, damaged sources, different poses, different formality levels.I thought it would be a weekend project — one system prompt and done. After 200+ test iterations I realized stable results require much more than prompt engineering. The main challenges:The AI subtlApr 3, 2026 6:32 AM
