AI如何拓展人们的工作内容How AI is expanding what people do at work

ServiceNow Secures Government AI Deals Driving 6% Stock Price Surge

ServiceNow (NYSE: NOW) experienced a 6% stock price increase following strong quarterly earnings and expanding public sector artificial intelligence (AI) contracts. Despite presenting a slightly conservative future revenue outlook for the next quarter, the enterprise software provider exceeded Wall Street expectations for its recent fiscal period. This strong financial performance, coupled with rising federal demand for automation, also lifted competitor Salesforce (NYSE: CRM) by 4%.

The growth is primarily driven by federal agencies accelerating their adoption of generative AI (GenAI) technologies to modernize legacy workflows. By integrating advanced machine learning (ML) capabilities directly into its proprietary Now Platform, ServiceNow allows government clients to automate complex administrative tasks securely. These high-value public sector contracts provide stable, long-term recurring revenue that mitigates softer commercial sector projections, positioning the firm as a highly efficient enterprise AI deployment vehicle.

This market response underscores a broader industry trend where tangible AI implementation, rather than speculative capability, drives enterprise software valuations. As government entities increasingly mandate secure, compliant AI integration, software vendors with established federal frameworks gain significant competitive advantages over legacy providers. The sector-wide valuation lift suggests robust institutional demand for practical, scalable enterprise AI solutions that deliver immediate operational efficiency.

Key Takeaways:
Government AI contracts are emerging as a major revenue driver for enterprise software companies. Source: Original Article
ServiceNow exceeded quarterly financial expectations despite issuing a conservative future revenue outlook. Source: Original Article
Analysts identify ServiceNow as a highly valued, cost-effective entry point for enterprise AI. Source: Original Article

Palantir ServiceNow Compete Enterprise AI Market 2 Distinct Software Architectures

Market analysis evaluates the enterprise artificial intelligence (AI) software market positioning of Palantir Technologies and ServiceNow. Due to limited information in the source citation, this evaluation focuses on their core architectural differences and market strategies. Palantir leverages its Artificial Intelligence Platform (AIP, Artificial Intelligence Platform) for deep data integration, while ServiceNow embeds generative AI (GenAI, Generative Artificial Intelligence) directly into enterprise workflow automation.

Palantir AIP enables enterprises to deploy large language models (LLMs, Large Language Models) within secure private data environments using its proprietary real-time ontology framework. Conversely, ServiceNow utilizes its Now Platform to automate IT service management (ITSM, IT Service Management) through its specialized Now Assist AI engine. Both companies report rapid enterprise adoption, with Palantir expanding commercial customer counts by over 40% and ServiceNow scaling high-value subscription revenues through AI-driven upsells.

This comparison highlights a broader industry trend where enterprise software providers must choose between deep data integration platforms or workflow-specific automation tools. Industry analysts monitor these distinct software architectures to determine which model yields higher long-term return on investment (ROI, Return on Investment) for corporate clients. The ongoing competition underscores the accelerating transition from experimental AI pilots to production-grade enterprise deployments across global business sectors.

Key Takeaways:
Palantir targets deep data integration while ServiceNow focuses on workflow automation.
Enterprise AI adoption drives significant revenue growth for both software platforms.
Secure LLM deployment remains a critical differentiator in enterprise software selection.

Source: Original Article

NinjaOne Partners ServiceNow 1 Unified Endpoint IT Management Integration

Based on limited information, endpoint management provider NinjaOne has partnered with ServiceNow to integrate their respective IT management platforms. This strategic collaboration aims to streamline IT service management (ITSM, IT Service Management) and endpoint management into a single, unified workflow. The integration allows IT helpdesk teams to access critical NinjaOne device data directly within the ServiceNow platform, effectively eliminating operational silos for global enterprises.

The technical integration enables bidirectional data synchronization, allowing helpdesk agents to view real-time device health, hardware specifications, and software inventory directly from ServiceNow incident tickets. This centralized visibility is expected to accelerate ticket resolution times by approximately 30% while minimizing manual data entry errors. Additionally, administrators can trigger remote troubleshooting actions, patch management, and software deployment tasks directly from the ServiceNow interface to improve operational efficiency.

This partnership highlights a broader industry trend toward consolidated IT operations and cloud-based security platforms. By bridging the gap between endpoint management and service desk workflows, both companies address the growing enterprise demand for automated management solutions. Organizations can leverage this unified system to optimize resource allocation, reduce operational overhead, and improve service delivery across highly distributed remote workforces.

Key Takeaways:
NinjaOne and ServiceNow integrate to unify endpoint management and IT service workflows.
Real-time device data visibility within ServiceNow reduces ticket resolution times.
The partnership reflects a growing industry demand for consolidated IT operations.

Source: Original Article

Google AI Overviews Reach 43 Percent Search Share Driving Conversational Queries

Google has rapidly expanded its AI-generated search feature, AI Overviews, which now appears in 43% of searches compared to 15% in the previous year. According to market intelligence firm Similarweb, monthly visits to Google's conversational AI Mode (AI Mode) grew from 126 million in June 2025 to 279 million by May 2026. This rapid adoption marks a major transition in how global web users access online information.

This technological shift changes Google from a portal of external links to a final destination that answers queries directly using indexed web content. Users now spend more time on the search platform, replacing short keyword searches with longer, natural conversational queries designed for AI processing. However, this development has severely reduced referral traffic for web publishers, as AI citations rarely result in direct user clicks to original sources.

To counter this loss of traffic, technology infrastructure provider Cloudflare has introduced tools that allow publishers to block AI crawlers. These security tools enable publishers to restrict content access unless AI companies pay for data licensing through specialized marketplaces. This trend highlights a growing industry conflict over intellectual property, web traffic distribution, and search engine monetization.

Key Takeaways:
Google AI Overviews now appear in 43% of searches, up from 15% last year.
Conversational AI Mode visits more than doubled, reaching 279 million in May 2026.
Publishers are losing referral traffic, prompting defensive measures like Cloudflare crawler blocks.

Source: Original Article

Moonshot AI Releases Kimi K3 3T Open Weights Long Horizon Coding

Moonshot AI has officially released the open weights for its Kimi K3 frontier model, making the system fully accessible for developer download and self-hosting. Previously restricted to API (Application Programming Interface) access, the Kimi K3 model is a 3T-class system designed for highly complex computational tasks. This landmark release allows developers to quantize, fine-tune, and integrate the model directly into their own local tools and platforms.

Moonshot AI designed the Kimi K3 model to specialize in complex tasks including long-horizon coding, repository-scale context management, tool use, web browsing, and multi-step planning. Due to its massive 3T-class scale, running the model locally requires substantial enterprise-grade hardware infrastructure rather than standard consumer setups. This open-weights release enables deep customization, allowing developers to perform low-latency quantization, offline deployment, and domain-specific fine-tuning for complex agentic workflows and proprietary applications.

The release reflects a growing global trend of developers adopting high-performance Chinese models like Kimi, GLM, and DeepSeek due to cost-efficiency and advanced capabilities. By providing open weights, Moonshot AI lowers the barrier for international enterprises seeking powerful alternatives to closed-source Western models. This strategic shift could accelerate the integration of open-source AI within mainstream software development pipelines.

Key Takeaways:
Moonshot AI democratizes advanced 3T-class model access through its open-weights release.
High hardware demands limit local execution of Kimi K3 to enterprise-grade setups.
Western developers increasingly adopt Chinese AI models for cost-efficiency and high performance.

Source: Original Article

AI Companies Shred Rare Books For High-Speed Dataset Digitization

Reports indicate that artificial intelligence (AI) development firms are purchasing and physically shredding rare books to facilitate high-speed digitization. Due to limited information in the primary source, the exact scale of this practice and the specific companies involved remain unconfirmed. However, this destructive scanning method aims to rapidly convert physical text into high-quality training datasets for large language models (LLMs).

Destructive scanning involves removing the bindings of physical books to feed individual pages into high-speed optical character recognition (OCR) scanners. This process significantly reduces digitization time and minimizes scanning errors compared to non-destructive manual page-turning methods. Consequently, AI developers can acquire clean, proprietary text data from out-of-print or rare historical volumes that are not yet available in public digital repositories.

This practice highlights the intensifying demand for high-quality, non-synthetic training data as AI developers face potential data depletion. It also raises significant ethical and preservation concerns within the academic and archival communities regarding the permanent loss of physical cultural heritage. As a result, the industry may face stricter regulatory scrutiny over data acquisition methods and intellectual property sourcing.

Key Takeaways:
AI developers use destructive scanning of rare books to accelerate dataset creation.
The practice highlights a growing shortage of high-quality digital training data.
Physical book destruction sparks ethical concerns regarding cultural heritage preservation.

Source: Original Article

Enigma Raises 70M Seed Launching 100 Robot Online Interaction Experiment

Robotics research startup Enigma has officially emerged from stealth after securing a 70 million dollar ($70M) seed funding round led by Index Ventures and Ribbit Capital. Sarah Guo of Conviction Partners also participated in the round, which will fund a large-scale public interaction experiment. The startup aims to use these real-time interaction insights to develop intuitive robotic control interfaces and alternative robotic brains.

The company developed its proprietary robotic arms and underlying artificial intelligence (AI) models entirely from the ground up to ensure tight hardware-software integration. Over 100 of these robots, housed in California and Israel, are now accessible online for real-time public interaction. This methodology prioritizes active human-to-machine engagement over traditional training methods like computer simulations, passive video observation, or human motion data collection.

Co-founded by former Microsoft researcher Jonathan Jacobi and cybersecurity expert Gal Niv, Enigma leverages technical expertise from Israel's elite Unit 8200. Their human-centric research model addresses a major bottleneck in AI robotics, specifically the training of foundation models (foundation models) for unprogrammed tasks. By analyzing direct human inputs rather than relying on pre-existing datasets, the startup aims to build highly intuitive interfaces and alternative robotic control systems.

Key Takeaways:
Enigma secured $70M in seed funding to pioneer human-centric robotic control interfaces.
A global online experiment allows public interaction with over 100 proprietary robots.
The startup bypasses traditional video-based training by focusing on direct human-to-machine engagement.

Source: Original Article

Retrieval Arena Launches Benchmarking 8 Metrics For RAG Pipelines

An independent developer has officially released Retrieval Arena, a new open-source benchmarking framework designed to quantitatively evaluate and optimize complex Retrieval-Augmented Generation (RAG) pipelines. The software tool systematically executes and compares various chunking strategies, retrievers, and rerankers against a unified golden evaluation dataset. By measuring performance across 8 distinct metrics, the framework successfully replaces intuitive design decisions with empirical, data-driven insights for AI software engineers.

The framework evaluates retrieval and generation quality as independent variables to prevent silent pipeline failures during production. It calculates precision, recall, Mean Reciprocal Rank (MRR), and Normalized Discounted Cumulative Gain (nDCG), while using an LLM judge to assess correctness and faithfulness. Initial testing on a 44-question dataset tracked latency and token costs per configuration, revealing that traditional BM25 retrieval outperformed pure vector search on the developer's specific document corpora.

This release highlights a growing industry trend toward the systematic optimization of complex Large Language Model (LLM) infrastructures. As enterprises deploy RAG systems, balancing retrieval accuracy against operational latency and token costs remains a primary engineering challenge. Standardized open-source evaluation frameworks like Retrieval Arena enable developers to diagnose critical bottlenecks, such as retrieved context being dropped due to budget constraints before generation occurs.

Key Takeaways:
Retrieval Arena benchmarks RAG pipelines using 8 distinct retrieval and generation metrics.
Independent evaluation of retrieval and generation prevents silent context-dropping errors.
Initial testing shows traditional BM25 retrieval can outperform pure vector search.

Source: Original Article

AI Integration Accelerates Software Vulnerability Discovery and Prioritization

Global cybersecurity priorities are shifting rapidly as artificial intelligence (AI) accelerates both software development and vulnerability discovery. While organizations initially leveraged AI to improve developer productivity, security leaders must now address the rapid escalation of software risk. Both malicious attackers and enterprise defenders are adopting these automated capabilities, making continuous security evolution an essential requirement across the modern software development lifecycle (SDLC).

Traditional security investments in vulnerability scanners, software composition analysis (SCA), and runtime monitoring provide necessary visibility but are no longer sufficient on their own. The primary operational challenge has transitioned from merely identifying software vulnerabilities to rapidly determining their exploitability and business impact. Consequently, organizations must deploy advanced risk-prioritization frameworks to identify which vulnerabilities pose the greatest threat to critical applications, allowing teams to remediate high-risk exposures first.

As AI continues to compress the timeline between vulnerability discovery and active exploitation, static security models are becoming obsolete. Organizations must transition to continuous, proactive defense postures that successfully align rapid development cycles with automated threat mitigation strategies. Ultimately, the future of enterprise cybersecurity depends on an organization's ability to make rapid, data-driven remediation decisions at scale.

Key Takeaways:
AI accelerates vulnerability discovery, forcing organizations to transition to continuous security models.
Visibility tools are insufficient without rapid exploitability and business impact prioritization.
Attackers leverage AI to speed up exploitation, requiring faster defensive remediation.

Source: Original Article

MultiAgent Healthcare AI Lacks 4Sector Coordination

Based on limited information from an incomplete industry report, modern healthcare systems are transitioning toward multi-agent artificial intelligence (AI) architectures. These emerging frameworks deploy 4 specialized AI agents to independently manage symptom assessment, clinical scheduling, insurance verification, and pharmacy coordination. However, while these domain-specific models successfully exchange basic data, they currently lack the capability to actively coordinate their operational activities to achieve unified patient care outcomes.

The primary technical bottleneck lies in the fragmentation of proprietary knowledge and distinct operational objectives across different machine learning (ML) models. Although these 4 specialized agents can transfer raw data, they lack the shared semantic understanding required for collective decision-making. Consequently, the absence of standardized coordination protocols prevents these autonomous systems from executing complex, interdependent clinical tasks, which currently necessitates human oversight to manage critical workflow transitions.

Resolving this coordination gap represents the next major milestone for clinical AI engineering and decentralized system architecture. Industry experts suggest that transitioning from basic data exchange to active agent collaboration could optimize healthcare delivery timelines by up to 40%. Developing robust, secure communication frameworks will be essential for deploying these multi-agent networks safely within highly regulated medical environments.

Key Takeaways:
Multi-agent healthcare AI systems currently lack the coordination required for autonomous workflows.
Standardizing communication protocols among 4 key clinical sectors will optimize operational efficiency.
Transitioning to collaborative AI agents could improve healthcare delivery timelines by 40%.

Source: Original Article

AI Drug Discovery Sector Faces 15 Year Development Cycles Under Eroom Law

The global pharmaceutical sector faces escalating financial and temporal pressures in drug discovery, which is increasingly defined by first-mover market advantages. Historical data since the 1950s reveals that the cost of developing new therapeutic compounds has approximately doubled every 9 years, a phenomenon known as Eroom's Law. Currently, bringing a single new drug to market requires an average timeline of 10 to 15 years and substantial capital investment.

These exponential cost increases and prolonged development cycles highlight the systemic inefficiencies within traditional pharmaceutical research and development (R&D) pipelines. AI-driven drug discovery platforms leverage machine learning (ML) to close the data loop, accelerating molecular screening and optimizing lead compounds. However, due to limited information in the provided source text, the specific technological implementations and precise modern cost figures cannot be fully detailed.

Overcoming Eroom's Law remains a critical objective for biotechnology firms seeking to establish sustainable and highly efficient innovation models. Transitioning from traditional trial-and-error methodologies to predictive, data-driven AI frameworks represents the most significant paradigm shift in modern pharmacology. This transition is expected to compress the 15-year development timeline and mitigate high clinical failure rates, though complete analysis remains restricted by limited information.

Key Takeaways:
Eroom's Law drives exponential cost doubling in pharmaceutical development every 9 years.
Traditional drug discovery requires up to 15 years, necessitating AI-driven pipeline acceleration.
AI integration aims to compress development timelines by closing the R&D data loop.

Source: Original Article

Enterprise Agentic AI Demands 6 Core Infrastructure Capabilities

Enterprise agentic Artificial Intelligence (AI) is rapidly transitioning from basic conversational chatbots to autonomous software agents designed to execute complex business tasks. These advanced agents must seamlessly operate end-to-end across human workers, organizational workflows, data repositories, and core systems. Due to the incomplete nature of the source text, this technical analysis is based on limited information regarding specific platform architectures and deployment strategies.

Deploying these autonomous agents requires a specialized infrastructure built to handle intensive computational and operational demands. The optimal enterprise platform must integrate key technical pillars, including sufficient Central Processing Unit (CPU) capacity, resilient data access, and policy-aware tool utilization. Furthermore, the architecture must support comprehensive observability, secure system integration, and advanced memory management to ensure stable execution. These technical requirements distinguish agentic AI platforms from standard Large Language Model (LLM) hosting environments.

This infrastructure shift highlights a major industry trend toward full workflow automation rather than simple information retrieval. As enterprises adopt agentic AI, upgrading underlying hardware, memory systems, and data pipelines becomes a critical operational priority. Ultimately, establishing these robust, policy-aware environments will determine the scalability, safety, and overall business value of autonomous agents in modern corporate settings.

Key Takeaways:
Enterprise agentic AI requires robust infrastructure beyond standard chatbot hosting capabilities.
Core platform requirements include CPU capacity, resilient data access, and memory management.
Successful deployment depends on policy-aware tool use and comprehensive system observability.

Source: Original Article

TechCrunch Disrupt 2026 Unveils Smart Systems Stage Addressing AI Grid Strain

TechCrunch has officially announced the Smart Systems Stage agenda for the upcoming TechCrunch Disrupt 2026 conference to address critical sector challenges. This specialized track will focus on the critical intersection of energy, physical infrastructure, and advanced technology. Due to limited information in the initial announcement, the primary focus centers on addressing the massive power demands and operational challenges of modern artificial intelligence systems.

The agenda highlights the critical utility grid strain that rapid AI expansion imposes on the global economy, focusing on emerging energy solutions like nuclear fusion breakthroughs. Industry experts will analyze the physical infrastructure upgrades, power distribution systems, and thermal management technologies required to sustain high-performance computing workloads. These sessions aim to bridge the gap between AI hardware capabilities and utility grid limitations, though specific technical specifications, speaker lists, and session times remain limited.

This event underscores a growing industry trend where energy availability directly dictates the pace of AI development and deployment. As computational models scale exponentially, energy infrastructure is rapidly becoming the primary bottleneck for technological progress. Consequently, technology companies must actively collaborate with energy providers and policymakers to secure sustainable, high-capacity power sources for future data centers.

Key Takeaways:
TechCrunch Disrupt 2026 will address the severe grid strain caused by AI infrastructure.
Fusion breakthroughs and energy innovations are critical to sustaining high-performance computing.
Energy availability has become the primary bottleneck for future AI model scaling.

Source: Original Article

IBM Integrates AI Governance With GRC Frameworks For Operational Risk

IBM has advocated for a fundamental shift in artificial intelligence (AI) governance, urging organizations to integrate AI risk management directly into existing enterprise governance, risk, and compliance (GRC) frameworks. Rather than managing AI models and agents in isolation, this unified approach addresses compliance, operational, and reputational liabilities. It connects AI risks directly to broader organizational operations to eliminate dangerous blind spots.

Modern enterprise AI deployments increasingly combine complex models, autonomous agents, tools, and Model Context Protocol (MCP) servers, which introduces significant operational risk. To manage this complexity, IBM proposes embedding key risk indicators (KRIs) and key performance indicators (KPIs) directly into active AI use cases. This integration allows business leaders to monitor performance and operational risk in real time, establishing clear, governed accountability across the enterprise.

As AI adoption accelerates globally, treating AI risk as an isolated category remains a significant operational and compliance liability for modern enterprises. Aligning AI governance with traditional operational risk management, third-party risk, and business continuity ensures a comprehensive view of enterprise risk posture. This systemic approach is essential for organizations to secure measurable return on investment (ROI) from their strategic AI developments.

Key Takeaways:
Siloed AI governance creates severe compliance, operational, and reputational liabilities for modern enterprises.
Integrating AI risk into GRC frameworks provides a unified view of operational risk.
Embedding KRIs and KPIs into AI use cases enables real-time performance tracking.

Source: Original Article