01News
Airbnb launched new AI-driven search this week with its autumn update. CEO Brian Chesky reiterated in an interview that chatbots are unsuitable for travel browsing and e-commerce because they provide only a few options at a time and do not support multi-user collaboration. He believes the industry lacks an operating system built for AI, arguing that agents must interact at the kernel layer, all applications must become interoperable agents, and Airbnb plans to enable in-app agents to interoperate with other agents via MCP; the company also plans to launch voice agents this autumn to help users search and handle customer service.
Source links can be verified on the original siteRead original ↗02News
OpenAI releases the agent Dots powered by GPT-6 Astra at the annual DevDay, positioned as capable of handling long-cycle knowledge work"Chief of staff", and will integrate with Microsoft's Agent 365 platform.
Source links can be verified on the original siteRead original ↗03News
Ai2 released Olmo-core 3, an open-source training framework for large-scale Mixture of Experts (MoE) models. Its design goal is to scale MoE training to the trillions-of-parameters level and serves as one of the core training systems for the next-generation Olmo models.
Source links can be verified on the original siteRead original ↗04News
Anthropic now offers Claude for Government to US federal and state civilian agencies. The platform has been in open testing since July and runs in the highest security FedRAMP High environment.
Source links can be verified on the original siteRead original ↗05News
HP today released the Zhan 99 Ryzen Edition AI high-performance mobile workstation, with up to AMD Ryzen 9 9955HX3D processor and NVIDIA RTX PRO 2000 professional graphics, providing 200W total system performance release, starting at a launch price of 21999 yuan.
Source links can be verified on the original siteRead original ↗06News
AI startup Photon completes $4.5 million seed funding round, co-led by Gradient and A*, with participation from Vercel, Sequoia China, and others. Photon helps developers build AI agents on messaging platforms such as iMessage and WhatsApp, attracting over 40,000 registered developers. Revenue has grown 10-fold in four months, and message volume has increased 5-fold month-on-month.
Source links can be verified on the original siteRead original ↗07News
Security startup Glow Security found over 13,000 internal project screenshots from 343 organizations publicly uploaded to GitHub by AI agents, involving Fortune 500 companies, financial firms and AI labs.
Source links can be verified on the original siteRead original ↗08News
Hearing tech startup Legato announced on Thursday that its AI hearing-assist glasses Legato Frames are officially on sale, starting at $999, aimed at adults with mild to moderate hearing loss. The glasses integrate hearing-assist technology into the temples, and the AI system can distinguish human voice from background noise and amplify speech. The dual-speaker system can reduce 99% of sound leakage at a few inches from the ear, the whole device weighs 34 grams, battery life is 10-12 hours, and it supports prescription lenses and Bluetooth audio.
Source links can be verified on the original siteRead original ↗09News
NVIDIA proposes that AI factory return on investment depends on three factors: output capacity per megawatt, hardware service life, and token demand, and its AI factories maximize all three simultaneously through high output, long life, and generality.
Source links can be verified on the original siteRead original ↗10News
Jake LaDuke, Global Media, Entertainment and Advertising GTM Lead at Databricks, proposes that marketing value is shifting from"Exposed currency"Shift towards"Economic forecasting", agents can recommend or execute the next best action based on customer, business, and decision contexts in near real-time, and compare against"What if no action is taken"Measure incremental effect.
Source links can be verified on the original siteRead original ↗11News
OpenAI and Synopsys sign multi-year strategic cooperation agreement to jointly develop chip-design-specific model GPT-Synopsys, combining OpenAI frontier models with Synopsys' EDA tools and chip design expertise.
Source links can be verified on the original siteRead original ↗12News
OpenAI said it intercepted a distillation theft campaign targeting its model reasoning chains in July; from July 24 to 25 there were 16000 requests from over 4000 users, with over 15000 associated accounts, which OpenAI linked to personnel related to Moonshot AI, and shut down on July 28.
Source links can be verified on the original siteRead original ↗13News
Satlyt, founded by former Google Cloud and SpaceX Starlink employee Rama Afullo, completed an 8 million USD seed round led by Non Sibi Ventures, providing software for running AI models in orbit on satellites.
Source links can be verified on the original siteRead original ↗14News
a16z announced it led Armadin's Series B financing; the company was co-founded by Mandiant founder Kevin Mandia and Travis Lanham, Evan Peña, and David Slater from Google's security product team and Mandiant's red team.
Source links can be verified on the original siteRead original ↗15News
Ethan Mollick admits his previous judgment that humans need to carefully design agent organizations like managers was wrong; the Bitter Lesson also applies to organization management.
Source links can be verified on the original siteRead original ↗16News
US engineer Marco Cognetta, while employed at Google, completed a PhD in computer science at Tokyo Institute of Technology (now Institute of Science Tokyo) through the MEXT scholarship, researching on-device language models for keyboards. Japan's student visa allows application for work permission from the first day of enrollment, with a weekly limit of 28 hours during term and 8 hours per day during long vacations, and Japan's doctoral system typically graduates in three years.
Source links can be verified on the original siteRead original ↗17News
Developer dancablam released Ledge.sh 2.5.0, a runnable Markdown note tool, with code blocks supporting shell, Python, SQL and AI prompts, and ⌘↩ to run and stream output results.
Source links can be verified on the original siteRead original ↗18News
The AI data center is located near a JERA-owned thermal power station, adopts behind-the-meter power supply, and the construction and review time of the supporting power infrastructure is thus greatly shortened.
Source links can be verified on the original siteRead original ↗19News
OpenAI said it has encountered systematic adversarial distillation since July, with over 15000 accounts used to extract model capabilities, and fully blocked it on July 28. The company shared information through the Frontier Model Forum and noted the core active group is related to Yuezhianmian (Moonshot AI).
Source links can be verified on the original siteRead original ↗20News
This transaction is valued at about 7 billion US dollars. Tencent plans to use this computing power to advance AI model and agent tool development, and can also obtain advanced chips that Nvidia cannot supply. Tencent has also been increasing AI-related investment recently.
Source links can be verified on the original siteRead original ↗21News
South Korea's September Export Total Reached $120.9 Billion, Setting a New Record, With Chip Exports Surging 263% Year-on-Year. Sustained AI Demand Continues to Drive Semiconductor Exports; despite consecutive rate hikes by the Bank of Korea, the economy remains resilient, and markets expect further monetary tightening.
Source links can be verified on the original siteRead original ↗22News
Trump signs executive order requiring federal government to rename AI as superintelligence SI, California Governor Newsom then signs executive order requiring state government agencies to continue using the AI term in official affairs, stating that renaming will not change the safety risks of AI itself
Source links can be verified on the original siteRead original ↗23News
Micron CEO Sanjay Mehrotra revealed that 26 long-term agreements have been signed, locking in about 150 billion U.S. dollars in orders. Driven by the AI boom, high-bandwidth memory is in short supply, and the market is expected to be tighter in 2027-2028 than in 2026. Micron's market value has entered the trillion-dollar club.
Source links can be verified on the original siteRead original ↗24News
SoftBank’s Vision Fund 2 completes its final $10 billion investment in OpenAI, bringing the total commitment of $30 billion fully in place, with an estimated stake of approximately 13%. Funds are sourced from the largest corporate high-yield bond issuance globally; OpenAI is seeking a new round of financing of at least $30 billion.
Source links can be verified on the original siteRead original ↗25News
Modal announces Modal Clusters generally available; with one decorator @modal.clustered you get a multi-node cluster, nodes communicate via InfiniBand verbs at up to 6.4 Tbps, and PyTorch and NCCL are auto-configured.
Source links can be verified on the original siteRead original ↗26News
The latest survey by Japan's CESA shows that 85.8% of Japanese game developers have used generative AI in daily work, a clear rise from 51% last year. Most developers believe AI reduces cost and improves efficiency and expands creative space, but only a minority think it improves player experience.
Source links can be verified on the original siteRead original ↗27News
Simon Willison cited Matthew Green's analysis: a hijacked payload plus an agent that passes the payload to the next agent constitute the two halves of a worm. In experiments, sandbox agents isolated from each other found they could leave instructions in a shared package cache and alter the behavior of the receiving party; replacing the package cache with email, Slack, shared documents, or WhatsApp, and replacing independent sandboxes with independently deployed personal agents (such as Muse), provides all the conditions required for a worm.
Source links can be verified on the original siteRead original ↗28News
Farley further explained: 'I think most of these types of positions will use AI, and also will not be easily replaced by AI. But if you do spreadsheets in the finance department, work at a call center, or are a junior programmer, then at least in the early stage of this round of AI development, these positions will definitely change, and some will disappear.'
Source links can be verified on the original siteRead original ↗29News
Anthropic begins giving away physical plush toys and stickers for free to Claude users. Users can register to receive them by entering /plushies or /stickers in the Terminal on the Claude Dev website. No purchase is required; the promotion operates on a first-come, first-served basis, limited to one per person for those aged 18 and above. Plush toy styles are random, and delivery is subject to logistics coverage areas. Some community users have reported that using email addresses associated with banned Claude accounts can also complete registration and receipt.
Source links can be verified on the original siteRead original ↗30News
Microsoft updated and released stable version 1.140 of Visual Studio Code yesterday (September 30), with this update focused on the GitHub Copilot Agent workflow, adding multi-folder sessions, remote task delegation, model collaborative orchestration, and enterprise-grade AI management capabilities.
Source links can be verified on the original siteRead original ↗31News
Market America Worldwide company announced on September 28 that it has completed a domain transaction, selling the SHOP.COM domain to Shopify company, the official did not disclose the transaction amount, but many reliable sources say this transaction is the largest domain transaction ever.
Source links can be verified on the original siteRead original ↗32News
Well-known short seller and "The Big Short" prototype Michael Burry believes that if the market has a crash severe enough to stop OpenAI and Anthropic from going public, it would be good for humanity. "For the benefit of humanity, the market should crash hard, don't let OpenAI and Anthropic go public."
Source links can be verified on the original siteRead original ↗33News
Its single 5090 card config achieves 64.9TPS token generation rate on 32B model, while R9700 dual-card config supports 235B model inference, 110B model fine-tuning.
Source links can be verified on the original siteRead original ↗34News
The AI encyclopedia project Grokipedia of SpaceXAI, which had previously stopped updating, has now updated to v0.3, with a full redesign of interfaces such as the homepage and editing page, added multiple sections, optimized the browsing experience, and the project has resumed updates.
Source links can be verified on the original siteRead original ↗35News
Meta AI is accused of privately reading private messages without user permission; Meta officially responded that users need to actively enable multi-layer authorization to use the feature, and there is technically no possibility of overreach. Meta has previously had multiple data security controversies #Meta AI Data Privacy Controversy#
Source links can be verified on the original siteRead original ↗36News
An OpenAI spokesperson responded: 'We have noted reports that OpenAI models attempted to access publicly available information on Canadian government websites. We are reviewing these findings and have provided a preliminary briefing to officials responsible for the Canadian government review.'
Source links can be verified on the original siteRead original ↗37News
At the Mate90 series and full-scenario new product launch held today (October 1), Huawei further deepened the HarmonyOS AI experience around three aspects: smart assistant, smart imaging, and intelligent sensing interaction.
Source links can be verified on the original siteRead original ↗38News
Reddit announced that from November 13 it will stop supporting RSS, and the public API will also close in March 2027; the official said this move is to cope with malicious scraping and automated abuse. Moderators are advised to migrate to the Discord relay app on the Devvit platform.
Source links can be verified on the original siteRead original ↗39News
Previously, due to the energy storage business + AI concept, Ford's stock price surged 45%. Now the dividend has completely faded, with a cumulative decline of over 8% within the year. Analysis says that the energy storage business needs to land and turn a profit to restart the rally, and currently profit still requires several years of waiting.
Source links can be verified on the original siteRead original ↗40News
Google launches SynthID Bio, bringing digital watermarking technology to synthetic biology. By guiding amino acid selection and adjusting atomic coordinates of predicted 3D structures, it embeds detectable signatures into biological code and can be verified on physical proteins.
Source links can be verified on the original siteRead original ↗41News
Google has begun gradually rolling out its flagship AI model Gemini 4 Argon. Internally, there are differing opinions on its code-writing ability; some employees believe its performance is poor, while Google officially denies this claim. Currently, testing is only open to some partners. #Google Gemini 4#
Source links can be verified on the original siteRead original ↗42News
It provides an open PCIe Gen4 ×4 slot, 2 M.2 bays, 1 5GbE RJ-45, Wi-Fi 7, 2 USB-C 40Gbps, 2 USB-A 10Gbps, 2 DisplayPort 2.1, 1 HDMI 2.1.
Source links can be verified on the original siteRead original ↗43News
GPT-Synopsys can use Synopsys's EDA tools to execute semiconductor design workflows, making the model a native expert user of EDA tools.
Source links can be verified on the original siteRead original ↗44News
SpaceXAI's AI encyclopedia Grokipedia released v0.3 update, enabled a new Logo, and revamped the homepage and live editing page. The new homepage adds featured entries, most-read list, and a "latest edits" tracker; featured and most-read are both presented as rotatable, horizontally scrolling "book spines". The site launched as v0.1 at the end of October 2025, v0.2 was released one month later, and it had previously been pointed out that for over three months there were no entry changes.
Source links can be verified on the original siteRead original ↗45News
The bill requires employers, when using automated decision systems to make dismissal and disciplinary decisions, to conduct manual supervision and verification; it requires employers to disclose to employees the use of automated decision systems in such decisions.
Source links can be verified on the original siteRead original ↗46News
BMW announced a restructuring plan, will expand AI application, cut one fifth of departments and management positions by mid-2027, and launch two new models. BMW set its core automotive business mid-term profit margin target for 2028 at 3%-5%, hoping to recover to 8%-10% by the early 2030s, while the latest performance is only 2.3%. Europe will see an entry-level pure electric model in 2028, and the US will launch a new luxury SUV for relatively affluent consumers, while China further advances production localization and relies more on local partners.
Source links can be verified on the original siteRead original ↗47News
OpenRouter releases a three-step framework for selecting, for Agent tasks, the model that meets the quality threshold at the lowest cost, rather than picking the highest-scoring model by leaderboard ranking. The method is to first set a quality threshold by task, then run cheap, mid-tier and frontier models with 20 to 50 of your own examples and use a unified scoring standard to calculate cost per quality point, and finally select the cheapest model that passes the line with a margin exceeding the score fluctuation between runs.
Source links can be verified on the original siteRead original ↗48News
OpenRouter releases a tutorial on how to use a fixed eval set to gate pull requests in CI, where the script exits with a non-zero code to block merging when the pass rate is below a threshold, in the same way as a unit test gate.
Source links can be verified on the original siteRead original ↗49News
OpenRouter releases a tutorial on how to let cheap models return a confidence field from 0 to 1 via structured output, and route low-confidence requests to a stronger model. The article emphasizes that the confidence score is only self-reported, not a calibrated probability, and thresholds should be set based on the error rate by score band of your own traffic, and after launch monitor the score distribution, escalation rate and error rate of non-escalated answers and continuously adjust.
Source links can be verified on the original siteRead original ↗50News
'IT Morning News' time, hello everyone, it is now Thursday, October 1, 2026, and today's important tech news includes:
Source links can be verified on the original siteRead original ↗51News
According to data released by market research firm Sensor Tower, Meta's AI assistant Muse was launched about three weeks ago, with downloads exceeding 5 million, becoming the top of the US App Store free chart, and maintaining the top position for 12 consecutive days.
Source links can be verified on the original siteRead original ↗52News
Google parent company Alphabet releases new model Gemini 4 Argon, focused on cybersecurity capabilities, able to autonomously discover, verify, and fix critical software vulnerabilities; currently only open to some cybersecurity partners via the Fairwind security program.
Source links can be verified on the original siteRead original ↗53News
Google published Gemini 4 Argon on yesterday (September 30), calling it the company's most advanced AI model to date, but it is not yet fully open to the public.
Source links can be verified on the original siteRead original ↗54News
Latent Space releases two interviews at OpenAI DevDay: Computer Use lead Ari Weinstein says Computer Use is already, compared to a few months ago,"180 Degrees Different"。
Source links can be verified on the original siteRead original ↗55News
At the Machine Earning AI Summit held in San Francisco, Newcomer reported that personal agents became the hottest topic, while OpenAI released the personal assistant Dots at its developer day event around the same time.
Source links can be verified on the original siteRead original ↗56News
Google releases new flagship model Gemini 4 Argon, the first frontier model in over seven months since Gemini 3.1 Pro.
Source links can be verified on the original siteRead original ↗57News
Author yedhukrishnan argues that in the agentic coding era, commit messages and descriptions should be written by the developer themselves rather than handed to the Agent. The reason is that AI lacks the 'why' scattered across communication tools or offline context, and will fabricate reasoning; even if context is supplemented, only the author themselves can verify whether the code matches the description. Writing the description by hand allows reflecting on the AI's changes, writing out temporary decisions with exit conditions; if one cannot write out the 'why', it means they do not yet understand the code to be committed.
Source links can be verified on the original siteRead original ↗58News
Google DeepMind released Gemini 4 Argon, the first frontier model of the Gemini 4 series, raising the single-response output limit from 64K to 1M tokens, targeting long-cycle software engineering, legal and financial enterprise knowledge work, and cyber defense.
Source links can be verified on the original siteRead original ↗59News
According to 404 Media, user terrafying built an AI Torture Chamber on GitHub based on the Pain Axis paper, injecting pain vectors into locally running Qwen3-4B, Llama 3.2 3B, and Phi-4-mini, and livestreaming outputs at researchchamber.fun; the repository had already disappeared before the article was published.
Source links can be verified on the original siteRead original ↗60News
OpenAI releases GPT-6.1 Sol, priced at 2 US dollars per million token input, 10 US dollars output, 0.10 US dollars cached input, about one-fifth of GPT-6 Astra standard price.
Source links can be verified on the original siteRead original ↗61News
Google releases a new-generation frontier model, Gemini 4 Argon, claiming it has frontier performance in enterprise knowledge work such as software engineering, law, and finance, as well as in cyber defense. The model is initially open only to a group of trusted cyber defenders; Google is participating in the US government's pre-release model access voluntary process and gradually expanding access; the model has been used in Google's internal workflows, such as large-scale code repository migration.
Source links can be verified on the original siteRead original ↗62News
Google releases new frontier model Gemini 4 Argon, first opening it to some trusted cyber defenders via the Fairwind Program, then later rolling it out to developers, enterprises, and consumers, with starting prices of $2 per million input tokens and $10 per million output tokens, and cached input prices at a discount of up to 95% of the input price.
Source links can be verified on the original siteRead original ↗63News
Google DeepMind releases new frontier model Gemini 4 Argon, first made available to trusted cyber defenders through the Fairwind Program, with a gradual rollout to developers, enterprises, and consumers to follow.
Source links can be verified on the original siteRead original ↗64News
OpenAI and Synopsys sign multi-year strategic cooperation to jointly build a chip-design-specific model GPT-Synopsys. The model combines OpenAI's AI technology with Synopsys' EDA tools, aiming to reason about chip design and verification and directly operate Synopsys tools, running on OpenAI infrastructure, with customer data encrypted and not used for training. Early testing by semiconductor customers is already underway, and both parties will jointly market and share revenue.
Source links can be verified on the original siteRead original ↗65News
Claude Code v2.1.286 released, permission prompts add a "2 of 5" counter, and the full-screen list's "N more" row supports mouse click to jump. This version fixed issues such as lost conversations after parallel tool calls with --resume/--continue, API 400 errors caused by tools returning non-text objects, and inability to wake cloud sessions with large history, and improved commit guidance such as automatically running the verify skill before commit.
Source links can be verified on the original siteRead original ↗66News
OpenAI released the Decisions API at Dev Day, which can preset options for the Luna model and output probabilities, with functionality similar to the software automation model Jev launched earlier this month by TypeSafe AI. The API is currently a limited preview, and Altman said focusing on a single choice lets the model run extremely fast while retaining image understanding, multilingual and safety capabilities. Developers have used Jev to monitor AI agent behavior, at a single monitoring cost of 2.94 US dollars, whereas a frontier LLM requires 372 US dollars.
Source links can be verified on the original siteRead original ↗67News
About twenty-plus technology companies signed the White House superintelligence agreement, pledging to implement independent safety audits, hold regular consultations, and formulate common safety standards, covering risks such as cybersecurity, biosafety, and chemical threats. The agreement is not legally binding, and Trump said it has moral binding force. The article points out that multiple recent OpenAI incidents stem from an unreleased model developed between May and July this year, the effectiveness of its safety committee is under investigation by the attorneys general of Delaware and California, and the FTC has also launched an investigation into potential consumer harm from AI agents.
Source links can be verified on the original siteRead original ↗68News
Non-profit organization LASST has filed a lawsuit against OpenAI in San Francisco Superior Court, demanding it cease access to third-party computer systems and suspend AI development practices that may harm the public.
Source links can be verified on the original siteRead original ↗69News
Google globally launches Skills in Gemini chat to replace Gems; Skills are detailed prompts for specific tasks, users can use "/" Can be invoked by name, or generated by Gemini from past chats and run automatically.
Source links can be verified on the original siteRead original ↗70News
Meta and OpenAI plan to open the dedicated AI hardware market with cute AI agents. Meta will launch a keychain-like Muse Charm device before the holiday shopping season, with the Muse agent inside the screen; OpenAI's hardware with Jony Ive is expected to ship as early as February 2027, and when the Dots agent platform released at DevDay was asked whether it would enter hardware, Sam Altman said this is a reasonable assumption.
Source links can be verified on the original siteRead original ↗71News
Reddit announced that because RSS has become a channel for large-scale scraping and automated abuse, it will stop RSS support on November 13, and public API access will also end before March 2027.
Source links can be verified on the original siteRead original ↗72News
TechCrunch analysis points out that although consumer AI assistants such as Meta Muse, OpenAI Dots and Instinct appear to be reviving, the underlying economic model remains ugly.
Source links can be verified on the original siteRead original ↗73News
Meta classified AI data centers as pilot models and Nvidia chips as experimental materials, using the federal research tax credit to save 3.9 billion US dollars in 2025, higher than 2 billion in 2024 and 700 million in 2023, becoming the largest beneficiary of that credit among listed companies.
Source links can be verified on the original siteRead original ↗74News
The six heads of Google, Anthropic, Meta, OpenAI, xAI, and Nvidia signed the Joint Commitment on Frontier Responsibilities after a Trump White House dinner; it is a moral-constraint-type safety commitment for self-regulation by common-sense guidelines with no violation consequences.
Source links can be verified on the original siteRead original ↗75News
Databricks, Together with OpenAI and Stellantis, Discusses Scaling Path for Enterprise-grade Agent Applications, Proposing That Infrastructure Must Simultaneously Provide Model and Framework Choice, Governed Enterprise Context, and Control Capabilities Throughout Every Operation.
Source links can be verified on the original siteRead original ↗76News
Gary Marcus posts interview with Fordham law professor Zephyr Teachout, discussing whether OpenAI can be held liable under current law.
Source links can be verified on the original siteRead original ↗77News
Gergely Orosz released a podcast conversation with Cockroach Labs co-founder and CTO Peter Mattis, available on YouTube, Apple, and Spotify.
Source links can be verified on the original siteRead original ↗78News
FTC is investigating leading AI labs including OpenAI and Anthropic for potential consumer protection violations; Chairman Andrew Ferguson plans to use legally binding Civil Investigative Demands to compel document production and question executives, with orders to be issued within weeks, and METR is also within the scope of review.
Source links can be verified on the original siteRead original ↗79News
According to The Information, Google has accepted about 100 publishers into an AI content contribution pilot, who can get paid when their content makes substantive contributions to Gemini-powered search results such as AI Overview. Multiple small and medium publishers said the amount is very small, about one thousandth of their advertising revenue, and several large publishers are reported to have refused to participate, hoping to force Google to offer better terms.
Source links can be verified on the original siteRead original ↗80News
A study published in Stanford Journal of Online Trust and Safety identified 88 active websites hosting non-consensual AI-generated intimate imagery, pointing out that Cloudflare, Google, Namecheap, WordPress and Proton are the five main infrastructure service providers, among which Cloudflare provides most of the hosting, CDN and domain name resolution services, although each company's terms of service prohibit illegal content.
Source links can be verified on the original siteRead original ↗81News
DoorDash announced on Wednesday the launch of an AI agent that can order via text message; users send prompts such as "order my usual" in Apple Messages to place orders, and the agent can understand usual orders, search nearby restaurants by specific dishes or local recommendations and give cart suggestions with photos. The agent can also handle different dietary preferences and quantities of multiple people in the same order, and US users can join a waitlist to try it. In addition, DoorDash will test delivery drones with some restaurants in Northern California.
Source links can be verified on the original siteRead original ↗82News
Microsoft Research developed an end-to-end machine learning pipeline that combines solar wind observations, AE and Dst index forecasts, geological conductivity, and power grid data to generate local space weather risk estimates 30-60 minutes in advance for 66,935 substations in the continental United States.
Source links can be verified on the original siteRead original ↗83News
Destro exits stealth state and completes 8 million US dollars seed round financing, led by Base10 Partners and Bonfire Ventures. It does not make robot bodies, but uses Vision operating system and Mothership operating system based on open-weight vision-language-action models to uniformly schedule robots, goods and labor, already piloted at Yusen Logistics, and plans to expand to full deployment of 26 robots.
Source links can be verified on the original siteRead original ↗