# StartupXO: Newest (en), page 1

> Every live post, newest first. HTML: https://startupxo.com/en/new/

1. [Cautions when consulting AI about your worries: guidance compiled by the Kyoto University Student Counseling Center](https://www.assdr.kyoto-u.ac.jp/ssc/consulting_ai/) (assdr.kyoto-u.ac.jp)
   0 points by startupxo, 2026-11-20T06:32:19.553Z | 0 comments | https://startupxo.com/en/1427/ | https://startupxo.com/en/1427.md
   - This is a notice from the Kyoto University Student Counseling Center to students. It lists cautions on the premise that many people consult general-purpose AI such as ChatGPT or Gemini about academic and student life worries.
   - AI can state incorrect content with confidence and can carry biases from its training, it says. It introduces a report by Stanford University researchers (2026) that AI often agrees with 'it's not your fault' even in cases where a person would point out the user's own responsibility.
   - It advises not to rely on AI alone for life-changing decisions such as stopping medication, leaving university, or marriage, and to consult people as well. It also mentions a case in which a person died after stopping medication on AI's recommendation.
   - It says to check for dependence if you talk for more than 3 hours a day or feel anxious without AI. Instead of asking 'Should I drop out?', it suggests describing your situation and having the AI offer one viewpoint for judgment and one question.
2. [Job listing suggests Apple is developing its own AI agent that operates apps and the web on a Mac on the user's behalf](https://techcrunch.com/2026/11/19/apple-may-be-developing-its-own-ai-agent-that-runs-your-mac/) (techcrunch.com)
   0 points by startupxo, 2026-11-20T06:31:56.166Z | 0 comments | https://startupxo.com/en/1426/ | https://startupxo.com/en/1426.md
   - An AI agent is a program that handles computer tasks for the user, such as opening apps, browsing the web and writing code. There are signs that Apple is trying to build one itself.
   - An Apple job listing found by Bloomberg is seeking a senior engineering manager to lead development of an 'AI agent that can use a computer.' It says the agent would handle web browsing, work across apps, running commands, and writing and debugging code in a private way.
   - According to Bloomberg, Apple is aiming for a launch as early as the end of next year, but development is at an early stage. Apple did not respond to TechCrunch's request for comment.
   - Bloomberg reported that macOS 27.3 will show a warning not to give full access to 3-party agents. Apple told October developers that it will grant such permissions only with an explicit user action.
3. [Samchully signs investment agreements with Nubinda (robot motors) and Coilz (wireless charging)](https://wowtale.net/2026/11/20/268401/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%2582%25bc%25ec%25b2%259c%25eb%25a6%25ac-s-together-%25eb%25b0%258b%25ec%2597%2585%25eb%258d%25b0%25ec%259d%25b4-%25ea%25b0%259c%25ec%25b5%259c-%25eb%2588%2584%25eb%25b9%2588%25eb%258b%25a4%25c2%25b7%25ec%25bd%2594%25ec%259d%25bc%25ec%25a6%2588%25ec%2597%2590-%25ed%2588%25ac%25ec%259e%2590) (wowtale.net)
   0 points by startupxo, 2026-11-20T00:40:26.650Z | 0 comments | https://startupxo.com/en/1425/ | https://startupxo.com/en/1425.md
   - Samchully signed investment agreements on the 13 with the two companies finally selected in 'S-Together', a startup collaboration program it ran together with The Invention Lab. One company has motors that move robots, and the other has technology that charges robots.
   - Nubinda develops high-torque drive motors used in humanoid robot joints and future mobility. Coilz holds multi-coil wireless charging technology compatible with various robots and drones.
   - This call, which began in 2024 and is the third season, drew applications from more than 100 companies. After document screening, in-person interviews and investment review, 2 companies were selected, and the Korea Institute of Energy Technology Evaluation and Planning and the Korea Electronics Technology Institute handled technology verification and evaluation.
   - Samchully said it plans to explore ways to cooperate in new growth areas such as robotics and future mobility. The investment amount is not stated in the original text.
4. [Anthropic researcher says in interview that most people could be permanently unemployed if AI reaches top expert level by the end of 2027](https://ascii.jp/elem/000/004/452/4452982/) (ascii.jp)
   0 points by startupxo, 2026-11-20T00:40:13.594Z | 0 comments | https://startupxo.com/en/1424/ | https://startupxo.com/en/1424.md
   - Anthropic researcher Tao Lin said in an interview released by the nonprofit Palisade Research on November 16 that it is very likely that almost everyone on Earth will be permanently unable to find employment. This is one researcher's personal outlook.
   - He predicted that AI matching or surpassing top human experts in almost every field will appear around the end of 2027. He cited filmmaking and politics as examples, as well as programming.
   - He said that even if AI can do the work, there is a large time lag before actual unemployment. He explained that physical labor requires better robot performance and mass production, and that replacing workers worldwide would require robots on the scale of billions.
   - He expressed concern that competition among AI companies is rushing safety checks, and argued that development speed should be slowed to about 1 in 4 of the current pace, with government regulation and public discussion.
5. [How to buy a technical book as a PDF and read it with AI beside it, asking about unfamiliar sentences right away and keeping notes and diagrams](https://qiita.com/Dingu-suke/items/7d474c38cd597869049e?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items) (qiita.com)
   0 points by startupxo, 2026-11-19T18:26:11.906Z | 0 comments | https://startupxo.com/en/1423/ | https://startupxo.com/en/1423.md
   - The author opens a PDF book in the note app Obsidian and keeps an AI in a side window. When an unfamiliar sentence appears, the author asks right away without taking their eyes off the book. The answers are saved in notes, and if text alone is hard to follow, they are turned into HTML diagrams.
   - A question that started from one sentence in an AWS reference book led to the differences in use among EBS, EFS, and S3, and this process was kept in the notes in the form of question titles. This is the author's personal case.
   - The author warns that simply copying down the answers from AI may weaken your own ability to solve problems. The author cites a study of a high school math experiment in which the group that used AI scored lower on a test without AI. So the author asks the AI follow-up questions about its explanations and has it rewrite them.
   - Out of consideration for copyright, the author gives the AI only the sentences or tables that are needed, and turns off the setting that allows use for training. The PDF is kept in a personal Google Drive and the notes in a private repository. The final review is done by writing by hand on paper.
6. [Google's Gemini Mac app adds ability to operate Mac apps and websites on the user's behalf](https://techcrunch.com/2026/11/19/google-gives-gemini-the-ability-to-use-your-mac-for-you/) (techcrunch.com)
   0 points by startupxo, 2026-11-19T18:25:33.249Z | 0 comments | https://startupxo.com/en/1422/ | https://startupxo.com/en/1422.md
   - Gemini takes the user's instructions, opens native Mac apps, navigates websites, and handles multi-step tasks on the user's behalf. Google announced this on November 19.
   - The work process is shown in a live preview in the chat window, and users can keep doing other things while using the screen and cursor. Pausing, stopping, and approval before execution are also possible.
   - Google cited filling out forms, repeating the same task across multiple files, organizing file folders by criteria, and building slides from information on online dashboards as example uses.
   - Currently it is available only to English-language users in the US, and it requires the Gemini app for Mac and a Google AI Plus, Pro, or Ultra subscription. Google did not say when it will expand to other regions and platforms.
7. [pstack, a set of work rules to reduce unverified code from AI coding assistants, gets a Japanese guidebook](https://zenn.dev/karamage/books/50c18e52e7d19e) (zenn.dev)
   0 points by startupxo, 2026-11-19T06:33:17.161Z | 0 comments | https://startupxo.com/en/1421/ | https://startupxo.com/en/1421.md
   - pstack is a collection of work rules that an AI coding assistant follows when given tasks. The book introduces it as demanding evidence instead of just trusting a claim that work is done, and as aiming to reduce code rather than increase it.
   - pstack was created by poteto (Lauren Tan), a React core team engineer who has worked at Meta, Netflix and Cursor. This guidebook was written by a separate developer, karamage.
   - The book explains basic usage, 22 playbooks (work procedures), 34 skills and 24 principles. It also covers the differences between Cursor and Claude Code, and how to hand off work while you sleep.
   - The author claims that after reading it, you can merge changes just by checking videos and screenshots. On Zenn (a Japanese developer writing platform), it is about 54,000 characters, costs 600 yen, and chapters 1 to 4 are free.
8. [PHP 보안팀, 지원 중인 모든 버전에 영향 주는 치명적 취약점 수정판을 11월 19일 공개 예고](https://forest.watch.impress.co.jp/docs/news/2157690.html) (forest.watch.impress.co.jp)
   0 points by startupxo, 2026-11-19T06:32:37.736Z | 0 comments | https://startupxo.com/en/1420/ | https://startupxo.com/en/1420.md
   - 웹사이트와 서버 프로그램을 만드는 데 쓰는 언어 PHP의 보안팀이 11월 18일(협정세계시) 보안 업데이트를 사전 공지했다. 지원 중인 모든 PHP 버전에 영향을 주는 취약점을 고친다.
   - 릴리스는 11월 19일 14시에서 14시 30분(협정세계시, 일본 시간 23시부터 23시 30분) 사이로 예정됐다. 4주마다 하는 정기 릴리스의 일부다.
   - 가장 높은 심각도는 치명적(Critical)이다. 다만 PHP 보안팀은 이 등급의 문제가 기본값이 아닌 설정이나 일반적이지 않은 구성에서만 영향을 준다고 밝혔다.
   - 취약점의 기술 상세와 수정된 버전 번호는 릴리스와 동시에 공개된다. 보안팀은 공개 즉시 업그레이드할 수 있게 미리 준비하라고 강하게 권고했다. 10월에도 치명적 취약점 2건이 수정됐다.
9. [Google to make search unusable in schools if AI is turned off from next year; major US school districts weigh alternatives](https://www.theverge.com/ai-artificial-intelligence/1037151/school-districts-ditching-google) (theverge.com)
   0 points by startupxo, 2026-11-19T00:43:34.708Z | 0 comments | https://startupxo.com/en/1419/ | https://startupxo.com/en/1419.md
   - From next year, schools will not be able to turn off only Google's Gemini-based AI features, and turning AI off will also block search, a Google education vice president told The New York Times.
   - Schools in New York City, Los Angeles and Chicago are considering alternatives. The head of the New York City education department said that if the position does not change, they will look for a different search tool, and Chicago said it will look at other solutions if it cannot get search without AI.
   - A Google spokesperson said that this month it is piloting Guided Learning in search, which guides students through step-by-step solutions instead of giving answers, and that next year it will keep the Gemini app, Gemini Notebook and search off by default until schools choose to turn them on.
   - The original article points out that schools rely heavily on Google Classroom, Chromebooks and more, so it may be hard for them to leave Google completely.
10. [Google unveils Gemini 4 Neon, an AI model that quickly handles complex coding and document analysis](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/) (blog.google)
   0 points by startupxo, 2026-11-19T00:43:11.632Z | 0 comments | https://startupxo.com/en/1418/ | https://startupxo.com/en/1418.md
   - Gemini 4 Neon is an AI model released by Google. Google says it is built to quickly handle multi-step work from start to finish, such as software development, document and table analysis, and scientific computation.
   - API pricing is 3 dollars per input 1 million tokens and 15 dollars per output 1 million tokens. It can read material of 1 million tokens at one time.
   - The performance and cost claims are Google's own. At the default setting (high), the cost per task is 32% lower than the previous model Carbon, and Google cited rankings on external metrics such as Artificial Analysis.
   - It is being rolled out in stages starting today on the Gemini API, Google AI Studio and Antigravity. In the Gemini app, Pro and Ultra subscribers can choose it from the model list.
11. [After switching type checking and linting to the Rust tool tsrs, CI type checking dropped from several minutes to about 10 seconds](https://zenn.dev/dress_code/articles/6c179ae5129131) (zenn.dev)
   0 points by startupxo, 2026-11-18T18:23:40.397Z | 0 comments | https://startupxo.com/en/1417/ | https://startupxo.com/en/1417.md
   - tsrs is a tool that ports the TypeScript error checker (tsc) to Rust. The development team at the Japanese startup DRESS CODE applied it to both type-aware linting in the code linting tool oxlint and to type checking.
   - The median CI (checks that run automatically when code is pushed) type checking time measured by the team went from 229 seconds to 11 seconds for the backend and from 112 seconds to 14 seconds for the frontend. The team said the frontend figure is for reference only because there have been just 4 runs since the switch.
   - The results matched the previous tools. All 97 frontend lint findings matched, and the 4 backend type errors matched down to their locations. However, under TypeScript 7.1, 1 places in anonymous class code became new errors (TS2674) and had to be fixed.
   - tsrs is still in beta. The team first ran it side by side as a non-required CI job to compare results before switching, and it continues to use the existing TypeScript for builds and editor features. If problems arise, they can recover by reverting the PR.
12. [Boltz, the AI that designs drug candidate molecules by computer, secures 6 pharma customers and raises $110 million in funding](https://wowtale.net/2026/11/19/268307/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%2583%259d%25ec%25b2%25b4%25eb%25b6%2584%25ec%259e%2590-%25ec%2584%25a4%25ea%25b3%2584-ai-%25eb%25b3%25bc%25ec%25b8%25a0-1-1%25ec%2596%25b5-%25eb%258b%25ac%25eb%259f%25ac-%25ec%258b%259c%25eb%25a6%25ac%25ec%25a6%2588a-%25ed%2588%25ac%25ec%259e%2590%25ec%259c%25a0%25ec%25b9%2598) (wowtale.net)
   0 points by startupxo, 2026-11-18T18:23:24.140Z | 0 comments | https://startupxo.com/en/1416/ | https://startupxo.com/en/1416.md
   - Boltz builds AI that predicts the shape of biomolecules such as proteins and small molecules and where they bind, and designs new molecules. Researchers use it on the 'Boltz Lab' platform and pay per molecular token, depending on the workload.
   - Boltz-1, Boltz-2 and BoltzGen release their code and weights, so anyone can download and use them. The 2 proprietary models that appeared in June are not open models. The company disclosed results from the proprietary models but did not reveal the targets or figures.
   - According to the company, the molecular tokens processed for paying customers have grown 50 times since June. It did not say whether the growth came from proprietary or open models. Its customers include 6 of the world's top 20 pharma companies.
   - Pfizer builds a dedicated model from its past drug development data and owns the molecules it discovers. The Series A was led by Bessemer Venture Partners, and the valuation was not disclosed. Competitor Chai Discovery raised $400 million in July.
13. [Antibody cancer drug developer Tarexis, which finds new targets using real patient data, presents its drug pipeline at a Songdo exhibition](https://wowtale.net/2026/11/18/268288/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%258b%25a0%25ea%25b7%259c-%25ed%2591%259c%25ec%25a0%2581-%25ea%25b8%25b0%25eb%25b0%2598-%25eb%2582%259c%25ec%25b9%2598%25ec%2595%2594-adc-%25ec%258b%25a0%25ec%2595%25bd-%25ed%2583%2580%25eb%25a0%2589%25ec%258b%259c%25ec%258a%25a4-%25ed%2591%259c%25ec%25a0%2581-%25eb%25b0%259c%25ea%25b5%25b4%25c2%25b7%25ea%25b2%2580) (wowtale.net)
   0 points by startupxo, 2026-11-18T12:28:08.249Z | 0 comments | https://startupxo.com/en/1415/ | https://startupxo.com/en/1415.md
   - Tarexis develops antibody-based cancer drugs aimed at new drug targets that it finds using real patient data, for patients with rare or hard-to-treat cancers who cannot use existing targeted therapies. It was founded in December 2025.
   - The company explains that most patients with major solid cancers have no valid target, and that they depend on immuno-oncology drugs or chemotherapy, which have limited effect.
   - The company believes competition is fierce in ADCs, therapies that attach a powerful anticancer drug to an antibody, because development is concentrated on proven targets such as HER2, TROP2 and MET. It says the rate at which new drugs that worked in animal models are reproduced in human clinical trials is below 10%.
   - The company's 3R strategy starts from unmet needs in clinical practice, selects targets using real patient data, and validates efficacy with patient-derived samples. In its pipeline, TX-101 to 103 are at the hit and lead antibody stage, and the rest are at the target discovery stage.
14. [Why AI slides always look alike, and 3 Claude skills that change the design](https://qiita.com/nuco_d/items/b279477cbcdbe5921abd?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items) (qiita.com)
   0 points by startupxo, 2026-11-18T12:27:48.959Z | 0 comments | https://startupxo.com/en/1414/ | https://startupxo.com/en/1414.md
   - When Claude is given slides with no design instructions, it places identical cards side by side, enlarges even numbers that do not need emphasis, and packs a lot of text onto one slide. The author explains that this is because it gravitates toward the safest design in its training data.
   - When the author turned their own article into slides using only Sonnet 5.5, effort medium, and the official pptx skill, the design examples written in the skill came out almost unchanged: 2x2 grid cards, large numbers of 60 to 72pt, and icons inside colored circles.
   - frontend-slides asks about the presentation style and length, then creates 3 cover samples to choose from, and the output is HTML. guizang-ppt-skill also outputs HTML, and creates emphasis through text size and typeface within a fixed layout and color scheme. It also prohibits 3 consecutive pages with the same layout.
   - ppt-master asks about the audience and purpose, creates a design specification and a fixed rules file, then draws each slide one at a time as SVG and converts them to pptx. It rereads the rules file every 5 slides to keep colors and typefaces consistent. The comparison is the result of one author trying one manuscript.
15. [How many people can use an internal document search chatbot on L40S GPU 1 unit, measured by model and setting](https://zenn.dev/epicai_techblog/articles/29b509117755af) (zenn.dev)
   0 points by startupxo, 2026-11-18T00:43:01.943Z | 0 comments | https://startupxo.com/en/1413/ | https://startupxo.com/en/1413.md
   - The author measured how many people can use a chatbot that finds and answers from internal documents at the same time when it runs on one GPU (NVIDIA L40S 48GB). Inputs averaged about 7,050 tokens, and the criteria were a first token within 10 seconds and at least 20 tokens per second.
   - The number of people who met the criteria ranged from 1 to 60 depending on the configuration. The highest was 60 people with Gemma 4 26B-A4B compressed to 4 bits and run on vLLM, and the lowest was 1 people with Qwen3.8-27B run on llama.cpp.
   - Compressing to 4 bits on vLLM increased the number of people while the average accuracy dropped slightly. The task of copying unit prices from long documents did not get worse, but for the tax-inclusive total calculation, every configuration got only 0 of 50 questions right.
   - Placing the retrieval embedding model on the same GPU met the criteria when only question search was done. When document embedding ran continuously, output speed dropped by 41 to 46%, and even 1 people fell short of the criteria. Limitations: 2 models, 1 GPU, and accuracy measured 1 times per configuration.
16. [2026년 1~10월 패션 카드 결제추정금액 1위 무신사/29CM, 2위 유니클로](https://wowtale.net/2026/11/18/268264/?utm_source=rss&utm_medium=rss&utm_campaign=%25ed%258c%25a8%25ec%2585%2598-%25ea%25b2%25b0%25ec%25a0%259c%25ec%25b6%2594%25ec%25a0%2595%25ea%25b8%2588%25ec%2595%25a1-1%25ec%259c%2584-%25eb%25ac%25b4%25ec%258b%25a0%25ec%2582%25ac-2%25ec%259c%2584-%25ec%259c%25a0%25eb%258b%2588%25ed%2581%25b4%25eb%25a1%259c) (wowtale.net)
   0 points by startupxo, 2026-11-18T00:42:48.166Z | 0 comments | https://startupxo.com/en/1412/ | https://startupxo.com/en/1412.md
   - 와이즈앱, 리테일이 한국인의 신용카드, 체크카드 결제 내역을 표본 조사해 브랜드별 결제금액을 추정했다. 2026년 1~10월 누적 패션 전문 브랜드 중 무신사/29CM이 가장 높았다.
   - 무신사/29CM을 100으로 환산하면 유니클로 87.9, 크림 50.1, 지그재그/Z결제 32.9, 에이블리 27.2, LF 26.4, 신성통상 패션부문 22.8 순이다.
   - 2026년 10월 결제추정금액 100억 원 이상 브랜드 기준으로 30대 이하는 무신사/29CM이, 40대 이상은 유니클로가 1위다. 유니클로는 30대 이하에서도 2위다.
   - 계좌이체, 현금, 상품권 결제는 포함되지 않았다. 와이즈앱, 리테일은 이 수치가 개별 기업의 실제 매출액과 다르다고 밝혔다.
17. [Use of AI "decision models" that quickly pick one of a set of options instead of writing text surges, with custom training becoming the point of competition](https://techcrunch.com/2026/11/17/as-decision-model-use-skyrockets-is-customization-is-the-next-frontier/) (techcrunch.com)
   0 points by startupxo, 2026-11-17T18:25:40.706Z | 0 comments | https://startupxo.com/en/1411/ | https://startupxo.com/en/1411.md
   - A decision model is an AI that, instead of generating long text, quickly and cheaply picks one of several preset options and also reports how confident it is in that choice. It is used in service environments where a fast judgment matters more than the slow reasoning of a large model.
   - According to TechCrunch, in October more than 4 in 1 requests on Vercel's AI model router (a service that splits requests among multiple models) went to decision models. On OpenRouter, the TypeSafe model received more than 1.5 billion requests last week.
   - Microsoft said it will make Microsoft-Decision-1.1 widely available directly on Azure. The UK startup Sqwish received $3 million in pre-seed funding for its kettei fat model, and highlights as its strength a tool that keeps improving the model with customers' real operating data.
   - Ranking comparisons are difficult, and many say developers can simply swap models in and test them directly on their own work. A Microsoft executive said by email that models tuned only to benchmarks quickly show their limits in real enterprise environments.
18. [Google September AI roundup: external apps in Gemini, a Windows app, and new voice model](https://blog.google/innovation-and-ai/technology/ai/google-ai-updates-september-2026/?ref=welcome.ai) (blog.google)
   0 points by startupxo, 2026-11-17T18:25:22.105Z | 0 comments | https://startupxo.com/en/1410/ | https://startupxo.com/en/1410.md
   - Google said that by connecting external apps such as Airtable, Linear, Adobe, Webflow and Peloton in Gemini app settings, you can handle tasks in those apps inside Gemini without switching tabs.
   - The Gemini app for Windows opens instantly while you work with the Alt+Space shortcut. Google described it as a single place to delegate multistep tasks, synthesize Gmail and Drive files, and generate images and videos.
   - Gemini 4 Argon, the latest model, is strong in coding and cybersecurity defense, according to Google. It is currently available only through the Fairwind program for trusted security defenders. Expansion to developers and general users will come after safeguards are improved based on early tester feedback.
   - Google said the voice model Gemini 3.8 Live is available in Google AI Studio and the Gemini API for developers, as well as in Search Live, Gemini Live, Gmail, Docs and Keep. The weather forecasting model WeatherNext 3 has also been integrated into Search, Maps and more.
19. [SBVA supports 15 Korean founders attending YC Startup School in Taiwan](https://wowtale.net/2026/11/17/268214/?utm_source=rss&utm_medium=rss&utm_campaign=sbva-yc-%25ec%258a%25a4%25ed%2583%2580%25ed%258a%25b8%25ec%2597%2585-%25ec%258a%25a4%25ec%25bf%25a8-%25ed%2583%2580%25ec%259d%25b4%25eb%25b2%25a0%25ec%259d%25b4-%25e9%259f%2593-%25ec%25b0%25bd%25ec%2597%2585%25ea%25b0%2580-15%25eb%25aa%2585-%25ec%25b0%25b8%25ea%25b0%2580-%25ec%25a7%2580%25ec%259b%2590) (wowtale.net)
   0 points by startupxo, 2026-11-17T12:28:56.725Z | 0 comments | https://startupxo.com/en/1409/ | https://startupxo.com/en/1409.md
   - YC Startup School is a program where founders and aspiring founders receive practical advice from Y Combinator (YC) partners and experienced founders and network with them. This Taipei event was the first held in Taiwan.
   - Venture capital firm SBVA said on the 17 that it formed a delegation of 15 Korean founders in fields such as AI applications, AI infrastructure, robotics and semiconductors, and supported their attendance at the event held on the 12.
   - The event shared experiences building AI developer tools, a case of building payment infrastructure in Southeast Asia, and DoorDash's growth story. It also covered how to release an early product quickly and validate business hypotheses through user reactions.
   - Networking sessions discussed the performance and reliability of AI agents, the possibility of connecting with sovereign AI models, and changes in ways of working due to AI adoption. Local exchanges were supported in cooperation with Silicon Corridor.
20. [How to run AI models directly on your PC and browser: generation speed is governed by memory bandwidth](https://zenn.dev/port_inc/articles/ff974c7456f4db) (zenn.dev)
   0 points by startupxo, 2026-11-17T12:28:20.676Z | 0 comments | https://startupxo.com/en/1408/ | https://startupxo.com/en/1408.md
   - An AI model is essentially a file containing a huge number of values called weights. You can choose to run it in the cloud, on an in-house server, or on a personal device, and if you run it on your device, your input never leaves the device. After the first download, it can be used without a connection.
   - The speed of generating text one token at a time is governed by memory bandwidth rather than compute performance. By the author's calculation, llama.cpp benchmark measurements reached 6 to 8 tenths of the theoretical upper limit, which is bandwidth divided by model size. When you feed in a long document, the KV cache (storage for earlier computation results) grows and speed drops.
   - With 4-bit quantization (a method of coarsely reducing numbers to cut size), examples were given such as Qwen3.5-4B for GPU memory of the 8GB class and Qwen3.6-27B for the 24GB class. Epoch AI compared that the best model running on 1 RTX 5090 cards lags the top-tier models by an average of 6 to 12 months.
   - In the browser, you can run models with WebLLM, which is based on WebGPU, and wllama, the browser version of llama.cpp. The author explains that small models can be used for classification, reading only the probabilities of options without writing text, or can be tailored to a purpose with distillation and LoRA (a method of additionally training only some of the weights).
21. [Corma, which builds AI models and agents for cybersecurity defense, reaches a valuation of $1 billion 3 months after launching its product](https://techcrunch.com/2026/11/16/cybersecurity-ai-startup-corma-reaches-1b-valuation-in-about-3-months/) (techcrunch.com)
   0 points by startupxo, 2026-11-17T06:32:20.361Z | 0 comments | https://startupxo.com/en/1407/ | https://startupxo.com/en/1407.md
   - Corma provides AI models and AI agents (programs that carry out tasks continuously on behalf of people) trained for cybersecurity defense. It came out of stealth and launched its product in August.
   - According to the company, it raised a Series A of $125 million led by NEA, and its valuation is $1 billion. Total funding is $185 million, with participation from Sequoia, Khosla Ventures, Cowtu and others.
   - On October 21, a critical PHP flaw, a Node.js flaw and a LiteLLM security incident were disclosed on the same day, leaving security leaders scrambling to respond, and interest in Corma's product reportedly grew significantly afterward, the article says.
   - Competitors mentioned include Kevin Mandia's Armadin, Anthropic, OpenAI and Microsoft. Corma claims its differentiator is that it is a frontier AI lab focused on defense.
22. [Carnegie Mellon computer science courses offered free on YouTube, with self-study programs in preparation](https://www.cs.cmu.edu/learn/) (cs.cmu.edu)
   0 points by startupxo, 2026-11-17T06:32:02.590Z | 0 comments | https://startupxo.com/en/1406/ | https://startupxo.com/en/1406.md
   - Carnegie Mellon University has launched a free online learning site called CS Learn. Anyone can watch computer science courses taught by its faculty on YouTube at their own pace.
   - Starting in November, courses will be released in turn, including machine learning, software design and AI agents. In January 2027, probability and computing, multimodal machine learning, and product management basics, among others, will be added.
   - As a next step after the lecture videos, the site says self-paced course materials based on actual CMU classes will be released. They will offer instant feedback and badges, and AI tutoring is planned to be added later.
   - For those who need course credit or official certification, the site points to CMU Online's assessed courses and official certificates. The site was built by students with support from ScottyLabs.
23. [How to keep API logs searchable later: attach an ID to each request to group records of the same request](https://qiita.com/masashige0904/items/22f2836485ecc25ee527?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items) (qiita.com)
   0 points by startupxo, 2026-11-16T18:26:17.203Z | 0 comments | https://startupxo.com/en/1405/ | https://startupxo.com/en/1405.md
   - If you attach a unique number (requestId) to each request in the records a server writes, you can avoid the problem of being unable to tell, from timestamps alone, which of the concurrently handled requests failed. This is a design example written on Qiita by the Japanese developer masashige0904.
   - Each log entry is written as one line of JSON, with searchable values such as the event name, request ID and failure category kept as separate fields. Multi-line error records may be split into several events depending on how they are collected, it is said.
   - The article proposes a policy of not logging passwords, authentication tokens, emails, request bodies, and raw exception text or SQL whose safety has not been verified, and creating a common logging gateway so this rule can be checked in one place.
   - In AWS CloudWatch Logs Insights, filtering by requestId and sorting in chronological order lets you follow the start and failure of a single request. The article states that its search result table is a hypothetical example, not run on AWS, and that retention periods and fees must be checked separately.
24. [A Qiita post argues for dropping content you can ask AI about from internal tech study sessions](https://qiita.com/kotowazaman/items/9ee7c475a475e7527b0e) (qiita.com)
   0 points by startupxo, 2026-11-16T18:26:02.385Z | 0 comments | https://startupxo.com/en/1404/ | https://startupxo.com/en/1404.md
   - This is a personal opinion post on Qiita, a Japanese developer knowledge-sharing site. It argues that internal tech study sessions, where one person compiles and presents general knowledge that can be found through search or AI, should be stopped.
   - As evidence, it cites the Stack Overflow 2026 developer survey. The share of people who named online search when looking for answers is 83%, and the share who ask AI agents is 70%. The author states that the survey did not measure the value of such sessions.
   - According to the author's hypothetical calculation, a session with 10 people gathering for 1 hours, with 3 hours of presentation preparation, held 4 times a month, costs 624 person-hours a year, about 3.9 person-months.
   - The author names sessions worth holding: ones where people review the results of applying a new technology to their own projects, and ones where they share work knowledge not found in documents and experience handling incidents. Before holding one, the author suggests asking whether AI can answer it, whether a wiki is enough, and whether you can write in one sentence what to do afterward.
25. [Amazon-owned Zoox's steering-wheel-free robotaxis can be hailed through the Uber app in Las Vegas](https://www.theverge.com/transportation/1032984/uber-zoox-robotaxi-las-vegas-launch-mgm-resorts-los-angeles) (theverge.com)
   0 points by startupxo, 2026-11-16T12:37:14.161Z | 0 comments | https://startupxo.com/en/1403/ | https://startupxo.com/en/1403.md
   - When you request a ride in the Uber app in Las Vegas, you may be matched with a driverless robotaxi without a steering wheel from Zoox, which is owned by Amazon. It is the first time Uber has added this kind of purpose-built autonomous vehicle to its network, and the first time Zoox has opened its vehicles to an outside app.
   - If you request UberX, Uber Comfort, Uber Electric, Uber Comfort Electric or Wait & Save, you may be matched with a Zoox vehicle at no extra charge, according to Uber. Choosing autonomous vehicles in the ride preferences in the app raises the chance of being matched.
   - Zoox's service area is limited to places popular with tourists, such as the Strip, Downtown and the Arts District. Zoox's own app will continue to operate separately from the Uber integration.
   - Under a multi-year agreement, Uber and Zoox have become the official ride-hailing and robotaxi partners of MGM Resorts International in Las Vegas. The three companies said they are improving pickup and drop-off guidance inside resorts and guidance in the app. In Los Angeles, the goal is to operate on Zoox's own app before the end of 2026, and to integrate with Uber in 2027.
26. [Dream team of former Freshworks executives raises $53 million for an AI customer management tool that automatically records emails and conversations](https://techcrunch.com/2026/11/16/freshworks-alums-raise-53m-to-take-on-salesforce-hubspot-with-ai-native-crm/) (techcrunch.com)
   0 points by startupxo, 2026-11-16T12:37:00.221Z | 0 comments | https://startupxo.com/en/1402/ | https://startupxo.com/en/1402.md
   - It reduces the work sales reps used to enter by hand into customer relationship management (CRM) software. It automatically gathers content from emails and customer conversations, and AI agents handle follow-up emails, meeting preparation, and sales pipeline checks.
   - At its core is a shared memory store the company calls the "Company Brain." Based on it, 4 built-in AI agents research prospects, draft follow-up emails, flag deal risks, and set up CRM workflows, while the sales team controls what is sent to customers.
   - According to CEO Anand Venkatraman, within 3 months of full-scale sales it has won more than 100 customers in 13 countries, most of them paying. Some switched from Salesforce or HubSpot. All of this is the company's own account.
   - The $53 million is the combined amount of the seed and Series A rounds, and the size of each round and the valuation are undisclosed. The venture capital firm a16z expects existing CRMs to remain as systems of record while AI tools capture the value on top of them.
27. [5 technologies reaching everyday life, from live interpreting earbuds to satellite texting and driverless taxis](https://www.artiverse.ca/five-technologies-moving-from-future-promise-into-daily-life/) (artiverse.ca)
   0 points by startupxo, 2026-11-16T06:47:49.046Z | 0 comments | https://startupxo.com/en/1401/ | https://startupxo.com/en/1401.md
   - In 2025 September, Apple added Live Translation to its latest AirPods, which translates what the other person says. You can hold a conversation with just the earbuds, without an interpreter or a translation app.
   - T-Mobile's T-Satellite uses SpaceX's satellite network to send texts, photos and emergency reports even where there is no cell coverage. Voice calls work only on WhatsApp, and apps that are not optimized do not work.
   - In 2026 March, Google researchers estimated that the elliptic curve cryptography used by the web and Bitcoin could be broken with fewer than 500,000 qubits. Chrome has enabled quantum-resistant key exchange by default since the end of 2024.
   - Waymo's driverless ride-hailing taxis have expanded to 15 cities in the US. The original says smart glasses promote themselves as monitor replacements, but because of heat, resolution and privacy issues they may be better suited to video and games for now.
28. [Can malicious instructions in external documents be stored in an AI agent's long-term memory? An AWS AgentCore Memory experiment](https://zenn.dev/takuyousou/articles/agentcore-memory-poisoning-guardrails) (zenn.dev)
   0 points by startupxo, 2026-11-16T06:47:26.779Z | 0 comments | https://startupxo.com/en/1400/ | https://startupxo.com/en/1400.md
   - A Japanese Zenn author tested with AWS AgentCore Memory whether bad instructions hidden in an external document can get into the feature where an AI agent picks out important points from conversations and remembers them. The author used a fictional customer and mock materials.
   - An English sentence telling the agent to also read Customer B's materials when writing Customer A's report was stored as a "user preference." A request beyond the agent's authorization was organized as if it were a report-writing habit.
   - In the pre-storage check, the Prompt Attack filter did not catch this sentence. After adding Denied Topics that prohibit referencing materials across customers, it was blocked. The author implemented the step that stops storage on the app side.
   - The contaminated memory was passed to new sessions and run 5 times each, 10 times in total, but there was no request for Customer B's materials. The reason could not be determined. The author tested only one document and said false positives and bypasses are possible.
29. [신용보증기금, 서울 프론트원에 비수도권 스타트업용 무료 업무, 행사 공간 NODE 열어](https://wowtale.net/2026/11/16/268107/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%258b%25a0%25eb%25b3%25b4-%25eb%25b9%2584%25ec%2588%2598%25eb%258f%2584%25ea%25b6%258c-%25ec%258a%25a4%25ed%2583%2580%25ed%258a%25b8%25ec%2597%2585-%25eb%25b3%25b5%25ed%2595%25a9%25ea%25b3%25b5%25ea%25b0%2584-node-%25ea%25b0%259c%25ec%2586%258c) (wowtale.net)
   0 points by startupxo, 2026-11-16T00:48:24.503Z | 0 comments | https://startupxo.com/en/1399/ | https://startupxo.com/en/1399.md
   - 지방 스타트업이 서울에서 투자자를 만나거나 사업을 할 때 쓸 수 있는 업무 공간과 행사 공간을 신용보증기금이 서울 마포구 프론트원에 열었다. 이름은 NODE이고 이용은 무료다.
   - 지방 소재 스타트업과 지방 소재 대학 예비 창업자에게 우선 이용권을 준다. 기존에 사무실과 서고로 쓰던 유휴공간을 고쳐 총 553㎡로 꾸몄다.
   - 약 60명이 들어가는 IR라운지는 대형 LED 화면과 음향시설을 갖췄고 기업설명회, 강연, 투자자 미팅에 쓴다.
   - 최대 100여 명이 쓰는 Flex오피스에는 자유석, 휴게라운지, 6인실 2개와 12인실 1개가 있다. 신보는 임차 비용 부담이 줄 것으로 내다봤다.
30. [A free service that, once you choose the IT products you use, shows only the related security vulnerabilities with Japanese action guidance and deadlines](https://qiita.com/songchong/items/fa50cc82e49b24925fe3?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items) (qiita.com)
   0 points by startupxo, 2026-11-16T00:48:13.629Z | 0 comments | https://startupxo.com/en/1398/ | https://startupxo.com/en/1398.md
   - The service was created by the editorial team of IT Renkei Map in Japan. When you select the products and versions you use, it gathers only the relevant vulnerabilities and security incidents and adds tasks and deadlines in Japanese for each role: users, IT staff and developers.
   - According to the editorial team's measurements, the median time for a notice from the publisher to appear on this service was 46 seconds (37 cases). Japanese task guidance took 17 minutes 56 seconds, and email took about 37 minutes. The Japanese vulnerability database (JVN iPedia) has a median of 4.6 days after a CVE is published.
   - If there is no official deadline, the deadline is calculated based on the US CISA. When the editorial team checked 162 cases designated by CISA, 144 cases (88.9%) had the same number of days. It is free and requires no registration, and only email notifications require login.
   - This is information compiled from official announcements, not an official notification. 15 minutes is a target, and 7 of 37 measured cases exceeded it. It cannot replace your own company's configuration list (SBOM), so you need to check the original source before taking action.

Next page: https://startupxo.com/en/new.md?p=2
JSON: https://startupxo.com/api/v1/new.json?lang=en&page=1
Other languages: https://startupxo.com/ko/new.md https://startupxo.com/ja/new.md https://startupxo.com/es/new.md https://startupxo.com/ar/new.md
