LeituraPesquisaRadarFramework de investimento
Entrar / Cadastrar
TudoTecnologiaProdutosInvestimentoBolsaOutros

Leitura

2026-10-29198 posts

Atualizar

HPE Selects ServiceNow for Five-Year Agreement to Transform Cloud & AI Support Services

Material
Verificado 2026-10-29 16:13 GMT+8

US technology company Hewlett Packard Enterprise (HPE) has selected ServiceNow as the foundation for the support services of its Cloud & AI business. According to a press release published on October 28, 2026, this five-year agreement enables HPE to deliver AI-powered support at scale.

HPE will utilize the ServiceNow AI Platform to create a single foundation for case and service management. Key planned implementations include the ServiceNow Super Intelligence Control Tower, a central control plane designed to discover, secure, govern, observe, and measure AI agents, models, and identities across support services. This tower will enable HPE to apply NIST and EU AI Act compliance frameworks out of the box and detect when an agent operates outside its permissions, stopping it in real time.

Additionally, HPE will deploy Technology Provider Service Management (TPSM), Customer Service Management (CSM), Field Service Management (FSM), Workflow Data Fabric (WDF), Hardware Asset Management (HAM), and Service Exchange. These modules aim to consolidate support requests, optimize technician dispatching, unify data across systems, and enable multi-party workflows.

Paul Fipps, President of Global Customer Operations at ServiceNow, stated that the move connects every case, dispatch, and AI workflow into one operating model to resolve cases faster and govern AI in real time. Mark Bakker, Executive Vice President and General Manager of Operations at HPE, noted that ServiceNow provides a single foundation to connect teams and streamline processes for a more seamless customer experience.

Fontes:newsroom.servicenow.com

Pesquisa

Samsung Electronics Reports Record Q3 Operating Profit of 107.4 Trillion Won, States 2027 HBM Capacity Is Sold Out

Tópico · 三星HBM产能与定价Estrutural
2026-10-29 13:28 GMT+8

South Korean tech giant Samsung Electronics released its third-quarter financial results on October 29, 2026. The report shows consolidated revenue of 195.3 trillion won and an operating profit of 107.4 trillion won. This figure makes Samsung the first global technology company to surpass 100 trillion won in quarterly operating profit, driven primarily by the explosive growth in artificial intelligence (AI) semiconductor demand.

In the Device Solutions (DS) division, Samsung achieved sales of 147.1 trillion won and an operating profit of 106.5 trillion won. The memory business alone saw sales surge 427% year-over-year to 140.7 trillion won. Notably, sales of fourth-generation High Bandwidth Memory (HBM4) more than tripled quarter-over-quarter, accounting for 60% of total HBM revenue, marking a significant improvement in Samsung's position within the high-end AI chip supply chain.

Addressing market concerns about capacity, Kim Jae-jun, Executive Vice President and Head of Strategic Marketing at Samsung's Memory Business Unit, stated that negotiations for 2027 HBM supply at target prices are complete and the capacity is currently "sold out." He further predicted that tight supply conditions would persist until 2028. Additionally, Samsung is negotiating new Long-Term Agreements (LTAs) with 10 major global customers, following similar contracts signed with 10 others previously. These new agreements extend beyond traditional hyperscalers and server manufacturers to include smartphone, automotive electronics, and networking device sectors.

Despite strong semiconductor performance, Samsung's non-memory businesses face challenges. The Device eXperience (DX) division, responsible for consumer electronics and mobile communications, recorded an operating loss of 80 billion won in the third quarter, marking its second consecutive quarterly deficit. Samsung noted that while new products like foldable phones boosted revenue, rising costs for mobile memory eroded margins. The company plans to address cost pressures next year by adjusting its product mix to focus on flagship models and premium TVs.

Regarding capital returns, Samsung announced a cash dividend of 4,601 won per share, totaling approximately 30 trillion won, and plans to allocate up to an additional 75 trillion won for shareholder returns, with specific measures to be finalized by the board in January next year.

Fontes:미래를 보는 창 - 전자신문

Pesquisa

Qodo Solutions Engineer Cole Gannaway: Using Software Graphs to Detect Cross-Repo Blast Radius in AI Code Review

Tópico · Qodo代码审查瓶颈Visão rápida
2026-10-29 15:00 GMT+8

The acceleration of AI code generation has shifted the engineering bottleneck from writing to reviewing. In the latest episode of the AI Engineer podcast, Cole Gannaway, a solutions engineer at Qodo, explores the specific challenges arising from this shift. He notes that even a minor change to a data model can quietly break web, mobile, CLI, and MCP clients if its cross-repository dependencies are not fully understood.

Gannaway proposes a method of visualizing these hidden dependencies by building a software graph of repositories, thereby revealing the 'blast radius' of code changes. Additionally, he discusses transforming team tribal knowledge and past PR comments into custom review rules, as well as the principle that the agent generating the code should not be the one grading it. For engineering teams grappling with the scaling of AI-assisted development, this episode offers practical perspectives on context management and automated review.

Fontes:youtube.com

Pesquisa

Tencent Targets Southeast Asia as Launch Pad for Global Rollout of AI Agent WorkBuddy, Announces Visa Partnership

Tópico · WorkBuddyMaterial
2026-10-29 15:30 GMT+8

Chinese tech giant Tencent Holdings is using Southeast Asia as a launch pad to take its “cost-effective” artificial intelligence tools global, fronted by its workplace assistant WorkBuddy, amid cutthroat competition at home.

According to Dowson Tong, CEO of Tencent’s Cloud and Smart Industries Group, the company was targeting existing regional customers of its cloud computing and big data services, particularly in markets such as Indonesia. Speaking at an opening speech on Thursday at the company’s annual Global Digital Ecosystem Summit in Shenzhen, Tong noted that the international push coincided with triple-digit growth in overseas demand for Tencent’s AI services over the past year, with global businesses showing “very high interest and demand” for Chinese AI offerings.

To keep pace with demand, Tencent said it was stepping up its overseas infrastructure expansion, having added or expanded data centre capacity across markets including Germany, Malaysia, Mexico, Japan and Saudi Arabia. The expansion strengthened Tencent Cloud’s footprint to cover clients from more than 150 countries and regions, Tong said.

Tencent released the international version of WorkBuddy in early July across 21 markets, including Hong Kong, Macau and Singapore. Jade Ding, head of overseas business growth for WorkBuddy, said on Wednesday that “the penetration rate of AI agents in everyday life and workplaces is extremely low” in the international market, shedding light on future potential. According to domestic research firm Analysys, WorkBuddy ranked as the most popular desktop workplace AI agent in mainland China in June with nearly 21 million users, leading ByteDance’s Trae IDE and Alibaba Group Holding’s QoderWork.

Tong acknowledged that Tencent’s overseas AI drive remained in an “investment phase” aimed at driving early user adoption rather than immediate commercial returns, with monetisation slated for a later stage. At the same time, computing power constraints continue to challenge the scale of China’s AI expansion, despite significantly increased capital expenditure on data centre infrastructure this year. “We are still somewhat supply-constrained in terms of bringing AI agent capabilities to the public,” Tong said, citing surging compute consumption from WorkBuddy and growing AI use across Tencent products, including the Xiaowei assistant in its super app WeChat.

Separately, Tencent on Thursday announced a strategic partnership with Visa, expanding collaboration across cross-border payments, agentic commerce and enterprise AI solutions, among other areas.

Fontes:South China Morning Post

Pesquisa

South Korean AI Robot Firm XayZ Unveils Modular Accessory Concepts for DEUX Robot Ahead of RoboWorld 2026

Tópico · XayZ DEUX机器人Visão rápida
2026-10-29 10:03 GMT+8

South Korean AI robotics company XayZ announced that it will unveil four expandable accessory module concept designs for its semi-humanoid service robot 'DEUX' at the 'RoboWorld 2026' exhibition, held from November 4 to 7 at KINTEX in Goyang, Gyeonggi Province. The design aims to keep basic functions on the robot body while allowing flexible addition of vision, audio, projection, or spatial awareness capabilities based on the workplace needs.

According to XayZ, these modules connect via a unified interface on the back of the robot. The vision camera module adds depth perception; the microphone/speaker module optimizes voice interaction range; the projector module displays information on floors or walls; and the spatial awareness module integrates an additional battery, 3D LiDAR, and cameras into a 'backpack' form factor. Based on visual SLAM technology, this backpack configuration simultaneously extends continuous operation time and expands spatial recognition range.

XayZ emphasized that despite the added hardware, the design maintains the approachability of a service robot, with the camera module shaped like a hat and the battery module like a backpack. The company plans to finalize the connection methods and specifications based on this concept. Additionally, XayZ will showcase three other robots based on its proprietary 'BrainX' framework, including a new barista robot named 'BarisBrew Max'.

Fontes:스타트업 생태계의 모든 것 \'플래텀(Platum)\'

Pesquisa

Rick Vanover, Vice President of Product Strategy at Data Protection Company Veeam: Zero Trust Should Extend to AI Agents, with Recovery Precision Down to the Change Unit

Tópico · VeeamMaterial
2026-10-29 15:00 GMT+8

AI tools resolved to be introduced last year already feel like outdated technology this year. However, Rick Vanover, Vice President of Product Strategy at Veeam Software, believes the real risk lies not in the models themselves—when an AI agent connects to enterprise data and infrastructure, model issues become data issues, ultimately evolving into a threat to the entire business.

He cites the personal information leak incident at South Korean streaming platform Tving as an example: attackers stole 'developer access keys,' gained unauthorized access to internal systems, and leaked information for approximately 39.54 million accounts.

Vanover points out that if the permission design for AI agents is improper, it may far exceed the scope required for any single employee to complete a task. Agents can publish content, deploy software, update records, and execute workflows across multiple systems. Human oversight remains necessary, but it cannot control machine-speed operations in real time.

More perplexingly, existing zero-trust systems are largely built around people and devices. Automated accounts, APIs, and agents granted permissions operate directly on sensitive data beyond human sight, forming so-called 'shadow agents'; these security gaps unrecognized by enterprises may continue to widen.

The first layer of response he proposes is to make data itself the primary control line: classify which data is sensitive, apply access policies, masking, tokenization, encryption, and data loss prevention rules, thereby limiting what agents can query, send externally, or modify. These controls cannot be bound to specific applications—controls must follow the data as it moves across application, API, and infrastructure boundaries.

The second layer is a shift in recovery strategy. Vanover argues that when AI agents make mistakes, simply rolling back the entire system is insufficient. While major intrusions or large-scale data damage indeed require full recovery, applying the same method to localized errors would erase work completed normally afterward. He proposes 'change-unit recovery': tracking specific configurations or transaction records modified by AI agents and rolling back only the problematic parts to their original state. Implementing this requires fine-grained change logs, information on inter-system dependencies, and protected recovery points that AI agents cannot modify or delete after the fact.

No system can prevent errors with 100% certainty. When excessive permissions, loose security nets, and overly fast processing speeds overlap, even a tiny mistake can instantly balloon into a major incident. Therefore, security planning must begin with the candid premise that 'errors can occur at any time.'

It should be noted that Vanover's arguments are architectural recommendations attributed to his professional judgment, not independently verified industry standards or product test reports. He did not provide quantitative implementation costs for change-unit recovery, specific toolchains, or failure rate comparisons for deployed cases.

Fontes:미래를 보는 창 - 전자신문

Pesquisa

South Korean MedTech Firm FlowNix and Asan Medical Center Sweep All Three Accuracy Categories in MICCAI 2026 Blood Flow MRI Challenge

Material
2026-10-29 10:00 GMT+8

South Korean medical imaging analysis software company FlowNix (플로우닉스), in collaboration with a research team from Asan Medical Center (서울아산병원), won first place in all three accuracy evaluation categories at the official “4D Flow MRI Reconstruction Challenge” hosted by the International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI 2026). The results were reported by Platum on October 29, 2026.

4D Flow MRI is a technology that visualizes blood flow within vessels during a heartbeat in three dimensions, used to assess risks of aortic aneurysms, stenosis, and dissection, and to plan surgeries. However, its high-resolution acquisition typically takes 30 to 60 minutes, limiting widespread clinical use. This year’s challenge focused on drastically reducing acquisition time: participating teams had to accurately restore original blood flow information using AI while reducing acquired data volume to one-tenth to one-fiftieth of the original.

A total of 264 participants registered, with 22 teams making the final rankings, including UCLA, ETH Zurich, and UCSF. The collaborative team comprising FlowNix, Professor Lee Jun-gu’s team from Asan Medical Center’s Biomedical Engineering Research Center, and Kangwon National University secured first place in all three categories: “high-precision reconstruction from sparse data,” “cross-hospital/disease generalization,” and “cross-body-part generalization.” Notably, in terms of generalization performance, although all teams trained only on aortic data, this collaborative team ranked first in four metrics—image similarity, error, magnitude of blood flow velocity, and direction error—when tested on hospitals, equipment, diseases, and different body parts such as cerebral, carotid, portal, and renal arteries that were not part of their training set.

FlowNix CEO Ha Ho-jin stated that the technology demonstrates the ability to maintain blood flow information accuracy while significantly shortening scan times, and plans to integrate it into their blood flow analysis platform to accelerate clinical adoption. Professor Lee Jun-gu of Asan Medical Center noted that they will further develop the technology for real-world clinical environments, enabling patients to receive accurate blood flow diagnoses through shorter examinations.

Fontes:스타트업 생태계의 모든 것 \'플래텀(Platum)\'

Pesquisa

Voice AI Cannot Replace Screens, Not Because Models Are Insufficiently Smart

Material
2026-10-29 15:03 GMT+8

The training approach of Dessa, the research team under Block, appears unconventional in the field of voice AI: it completely removes text, allowing the model to learn dialogue directly from sound itself. According to the Block team, models trained this way can listen and speak simultaneously, interrupt naturally, provide instant affirmations like "uh-huh" or "yes," and even run locally on a MacBook Pro equipped with an M4 Max chip. The logic behind Dessa stems from Jack Dorsey's intuition—since human infants learn to speak before they learn to read, why can't machines follow the same path? However, limited by undisclosed training details and model scale, it remains difficult for outsiders to assess the upper limits of this technical route.

Meanwhile, mainstream commercial voice models are taking a different path. xAI's Grok Voice 3.0 focuses on real-time end-to-end voice interaction, but its foundation still relies on the text reasoning capabilities of large language models; OpenAI's GPT-Live improves the naturalness of multilingual accents in the ChatGPT client and adds nine new voice tones in its API. The common optimization direction for these models revolves around an extremely subtle timeframe: half a second. Waiting half a second longer while typing on a keyboard hardly changes the nature of communication, but in a phone call, a half-second silence is enough to push the conversation to the opposite extreme.

Internal operational data disclosed by xAI shows that this voice system, in addition to handling massive amounts of in-car conversations for Tesla vehicles daily, also handles about 20,000 customer service calls for Starlink, with about 70% of issues resolved without needing to transfer to human agents. In-car spaces and customer service hotlines constitute the application scenarios where voice AI can currently exert its greatest influence.

However, even though voice models can now handle tasks far more complex than 'turn on the lights'—such as helping users send and receive text messages, make restaurant reservations, and listen to developers' verbal descriptions to invoke programming tools to modify code—screens always have their place. Typing on a keyboard and clicking a mouse cannot yet be shaken by voice.

This is not because the models are not smart enough, but rather constrained by the physical attributes of sound as an information medium. Sound is linear; it is particularly suitable for expressing a vague intention, such as an idea in programming or directing an Agent to arrange tasks. But in work production scenarios, it is not good at allowing people to review multiple parallel options simultaneously. When writing an article, one needs to see the paragraph structure at a glance and backtrack to modify it at any time; when writing code, one must stare at the screen to compare code differences and running results; even just booking a flight ticket, spreading the prices, departure/arrival times, and change/cancellation rules of several flights on the screen for a quick scan is far more efficient than listening to a voice assistant read them out one by one.

Fontes:爱范儿

Pesquisa

Samsung Electronics' Mobile and Appliance Units Swing to Operating Loss in Q3 Despite Revenue Growth

Tópico · 三星电子DX部门利润承压Material
2026-10-29 09:39 GMT+8

According to a report by ZDNet Korea, Samsung Electronics disclosed its Device eXperience (DX) division's performance for the third quarter of 2026. While the Mobile eXperience (MX) and Network business unit's revenue grew 11% year-over-year to 37.7 trillion KRW, and the Visual Display and Digital Appliances (VD/DA) unit maintained flat revenue at 13.9 trillion KRW, the entire DX division's operating profit swung from a 3.5 trillion KRW gain in the same period last year to an 800 billion KRW loss.

Samsung attributed this decline to increased raw material and logistics costs, which outweighed the sales boost from new models like foldable phones. Specifically, the MX division's operating result shifted from a 3.6 trillion KRW profit to a 600 billion KRW loss. The home appliance division's losses widened from 100 billion KRW to 200 billion KRW. In contrast, subsidiary Harman performed well, with revenue up 9% to 4.3 trillion KRW and operating profit up 50% to 600 billion KRW, partly aided by US tariff refunds.

Looking ahead to Q4, Samsung plans to drive sales through new products such as the Galaxy S26 series and foldables, while minimizing profitability risks through process efficiency improvements. For appliances, the company aims to strengthen domestic marketing and expand subscription and B2B revenues to mitigate ongoing cost pressures.

Fontes:zdnet.co.kr

Pesquisa

Hancom CEO Kim Yeon-su Targets 50% Overseas Revenue by 2030 as AI Agent Platform Nomadian Launches in US This December

Visão rápida
2026-10-29 10:26 GMT+8

Kim Yeon-su, CEO of South Korean software company Hancom, stated at the 'Nomadian Day' event on October 29 that the company aims to generate 50% of its total revenue from overseas markets by 2030, with an interim target of 40% by 2028. This strategic shift centers on its newly launched AI workflow platform, 'Nomadian'.

Nomadian allows users to assemble teams of AI agents to handle tasks, similar to managing a company. It provides 12 basic AI team members and expands functionality by connecting external services via MCP and API. Hancom plans to start a beta service in the United States in December 2026, followed by a global official release in March 2027.

To enter the US market, Hancom is collaborating with Amazon Web Services (AWS) on its go-to-market strategy, targeting sole proprietors and small businesses that already use or pay for AI services. The company identified an initial addressable market of 5.21 million enterprises. Standard pricing is expected to be around $40 per month, with an initial 50-60% discount planned to validate the pricing structure.

Fontes:zdnet.co.kr

Pesquisa

Anthropic's Claude Code Team Member Addy Osmani: You May Be Over-Prompting the New Version of Claude

Material
2026-10-29 14:13 GMT+8

Addy Osmani, a member of Anthropic's Claude Code team, published a long-form post on X (formerly Twitter) with the core assessment that newer versions of the Claude model require significantly less prompting than what most users are still providing. He cited a specific figure: the Claude Code team removed over 80% of the content from the system prompts for Claude Opus 5 and Claude Fable 5, resulting in no measurable degradation in coding evaluation benchmarks.

Based on this practice, Osmani’s operational advice is to give Claude a single prompt containing a clear goal and verification conditions (i.e., how the model should confirm the task is complete), and then let it work. Start with the /effort medium setting and adjust as needed.

For situations where you are not yet clear on exactly what needs to be done, he suggests first letting Claude interview you until it can draft a specification document. Then, start a new session and use that spec to build what you want. The key mechanism here is that if a task can be verified by a pass/fail result or a numerical target to beat, and the report format is readable by the model, Claude can continue working autonomously without your supervision.

Osmani provided an actionable audit method: when a new model is released (such as Opus 5.5), remove some of the content from your own instructions and only add back those elements that the model actually misses. The /doctor prompt-audit command in Claude Code flags instructions written for older models. He recommends auditing your workflow every few months to ensure you are not writing instructions based on how older models used to behave.

Regarding the use of subagents, Osmani’s judgment is quite clear: Claude will delegate tasks to subagents on its own, so users rarely need to request them explicitly. Only explicitly request subagents for large, independent tasks, such as long audits or migrations. For small tasks, subagents multiply costs and latency. His core view is that while subagents can parallelize work or provide a second opinion, they do not make outcomes smarter by default. For agents you build yourself, a single agent equipped with skills may outperform a subagent design in quality, often at lower cost and latency—a judgment he explicitly limits to personal observation.

It is important to note that the data regarding the 80% reduction in system prompts comes from the Claude Code team’s own internal practice reports; independent benchmark methodologies or specific evaluation sets have not been publicly disclosed. Osmani’s recommendations are largely based on his first-hand observations as a team member, representing attributable practical methods rather than peer-validated experimental conclusions.

Osmani also mentioned that Claude Code recently set auto mode as the default startup mode for interactive terminal sessions. In this mode, a classifier analyzes Claude’s behavior rather than relying on traditional permission prompts. He explicitly warned that this is not a guarantee of safe behavior—neither is blindly approving every prompt. When Claude reports completion, you should ask it to show you the commands, outputs, or screenshots.

Fontes:AIHOT

Pesquisa

Multiple IT and OT Vendors Disclose PHP2Shell Vulnerability Impact and Patch Status; Industrial Systems from Siemens and Schneider Affected

Tópico · PHP2Shell漏洞Material
2026-10-29 12:38 GMT+8

As the PHP2Shell vulnerabilities (CVE-2026-109618 and CVE-2026-109619) become public, major global Information Technology (IT) and Operational Technology (OT) vendors have released security advisories disclosing the impact on their products and remediation progress. Since PHP is widely used in management interfaces and backend services, this vulnerability threatens not only web servers but also industrial automation systems and critical infrastructure.

In the OT sector, Germany's Siemens confirmed that its RUGGEDCOM APE1808 series routers are affected if running specific versions of Palo Alto Networks PAN-OS, recommending users upgrade to available versions. France's Schneider Electric stated that EcoStruxure Security Expert versions prior to 1.48.4.0, all versions of ProLeiT LINEoptimize, and EcoStruxure Control Engineering are affected, providing corresponding upgrade or patch solutions. Switzerland's ABB indicated risks in its B&R APROL system and MES systems bundled with Dokuwiki, planning to release patches for specific versions by late October. US-based Johnson Controls also confirmed that its CEM AC2000 building automation controllers and certain video surveillance products are affected, offering hotfixes.

In the IT and security software domain, Alex Spokoiny, Chief Security & Trust Officer at Check Point, argued that enhanced AI capabilities have led to the disclosure of more severe vulnerabilities, presenting challenges similar to the previous Log4Shell incident. Check Point itself confirmed that Force Firewalls, Spark Firewalls, and Cloud Firewalls are affected when enabling specific remote access features, with hotfixes now available for most versions. Citrix announced that NetScaler ADC and Gateway are affected due to their PHP-based management interface, expecting to release patched versions on October 27. Broadcom, Cisco, and AWS also disclosed affected components and repair timelines for their respective product lines, with AWS noting that some Amazon Linux versions were affected but patched in late October.

Notably, some vendors such as Moxa, Nozomi Networks, and Rockwell Automation have explicitly declared that their products do not use PHP or are unaffected by this vulnerability. For enterprises relying on these industrial control systems, immediately auditing assets against detailed product lists published by each vendor and prioritizing patches for management interfaces exposed to the internet or internal networks is the key step to reducing remote code execution risks.

Fontes:iThome

Pesquisa
Próxima →
Entrar / Cadastrar
LeituraPesquisaRadarFramework de investimento

The greater the authority AI has to act on behalf of humans, the more humans often need a visual interface to verify what it has actually done, and to stop and retract in time before errors occur. This is a structural need unrelated to model capabilities: supervision requires parallel information, while sound can only be serial.

Physical environments and social etiquette also secretly define the boundaries of voice. When driving, cooking, or holding a child with both hands, vision and limbs are occupied, so speaking is undoubtedly the most effortless savior; but once the scene switches to a quiet open-plan office, a person talking continuously to a computer still seems too weird. These constraints concern information bandwidth and privacy boundaries, which no matter how many milliseconds engineers press down the model latency, cannot solve.

Dessa's 'non-literacy' route and Grok's 'half-second latency' optimization represent different explorations in voice AI regarding training philosophy and engineering implementation. They are both pushing the upper limits of voice interaction capabilities, but neither changes a basic fact: the linear nature of sound determines that in high-bandwidth, multi-option, parallel-review work scenarios, it remains a supplement to screens rather than a replacement. The more machines learn to speak like humans, the clearer we realize which moments allow speaking, and which moments require keeping quiet.

Overall, the transferable insight from this methodology is that while model capabilities are advancing, user workflows often lag behind. Regularly auditing prompts, trusting the model’s autonomous execution under clear verification conditions, and avoiding over-engineering subagents for small tasks are practical paths to reducing usage friction and costs.