AI Agent News
实时追踪 AI Agent 赛道的重大事件、融资动向、模型发布和技术突破
最新行业资讯
实时追踪 AI Agent 赛道的重大事件、融资动向、模型发布和技术突破
重大事件时间线
OpenClaw GitHub 爆发
OpenClaw 10 天冲上 GitHub 全球 Top 10，超越 Linux 内核 Star 增速
Meta 20亿收购 Manus
Meta 以 20 亿美元收购 Manus AI，通用 Agent 赛道正式被巨头锁定
DeepSeek-V3 开源
性价比之王，成本仅 GPT-4 的 5%
Manus 一夜爆火
全球首款通用 AI Agent 在国内社交平台引发空前关注
OpenAI Deep Research
OpenAI 推出深度研究 Agent，一键生成专业研究报告
MCP Server 破 500
MCP 生态爆发，3 个月构建 500+ Server
DeepSeek-R1 震惊全球
开源推理模型，成本仅 OpenAI 的 3%，引发全球 AI 格局震动
MCP 协议诞生
Anthropic 发布 Model Context Protocol，成为 Agent 接口事实标准
Claude Computer Use
Anthropic 让 AI 首次直接操控电脑屏幕，开创计算机使用新范式
Replit Agent 全栈自动化
自然语言到上线产品，面向非工程师
Cursor ARR 破亿
史上增长最快 SaaS，AI 编程工具新王者
Claude 3.5 登顶 SWE-bench
最强编程 AI，Bug 修复能力达到初级工程师水平
Devin 发布
全球首个自主 AI 软件工程师，能独立完成完整编程任务
AI Code Review Tools in 2025: CodeRabbit vs PR-Agent vs GitHub Copilot Compared
Engineering teams adopting AI code review report 15-30% fewer production bugs and 40% faster review cycles. Comparison of top tools across 500 PRs: CodeRabbit leads in security vulnerability detection (78%); Sourcegraph Cody excels at logic errors with codebase context (58%); PR-Agent has the lowest false positive rate (12%) and is the only open-source, self-hostable option. GitHub Copilot offers best integration for GitHub-first teams already on Copilot subscription ($19/month). ROI for a 20-person engineering team: typically 5-10x investment from reduced production bugs and faster reviews.
The AI Agent Economy: New Job Categories Emerging as Autonomous AI Scales
As AI agents handle increasing portions of knowledge work, new job categories are emerging: AI Workflow Orchestrators who design and manage multi-agent systems, AI Output Reviewers who quality-check and curate AI-generated work at scale, Prompt Architects who specialize in system design for complex AI workflows, and AI Ethicists who monitor and audit AI systems for bias and harm. LinkedIn reports 340% YoY growth in job postings mentioning "AI agents" and "LLM orchestration." Meanwhile, demand for traditional data entry, basic research, and routine writing roles has declined 25-35% year-over-year.
Salesforce AgentForce Reaches 5,000+ Enterprise Deployments in 6 Months
Salesforce reported AgentForce, its AI agent platform, has been deployed by over 5,000 enterprise customers within 6 months of launch—the fastest adoption curve in Salesforce history. Key metrics: 2.5 million AI agent actions executed daily, 3.5x average productivity improvement in target workflows. Top use cases: customer service automation (60-70% reduction in human agent handling time) and sales development automation (40% increase in meetings booked). AgentForce has contributed to AI-related revenue exceeding $1B ARR, with Einstein 1 platform becoming Salesforce's fastest-growing product line.
NVIDIA Blackwell B200 GPU大规模交付：AI训练成本降低70%
NVIDIA开始大规模交付Blackwell B200 GPU，相比H100训练速度提升4倍，推理吞吐量提升30倍。微软、亚马逊、谷歌、Meta均已部署数万张B200。配合NVLink 5.0和GB200 NVL72服务器，AI训练成本预计下降70%，将加速AGI研究和大规模AI部署。
EU AI Act Takes Effect: What Enterprises Need to Do Now
The EU AI Act, the world's first comprehensive AI regulation, has entered its enforcement phase for high-risk AI systems. Organizations using AI in employment decisions, credit scoring, medical devices, critical infrastructure, and education must now comply with documentation, transparency, human oversight, and accuracy requirements. Penalties reach €30M or 6% of global annual turnover—whichever is higher. Legal experts estimate 40% of enterprise AI deployments require modifications for compliance. Key deadlines: high-risk systems must be registered in EU database by Q4 2025.
微软Copilot Studio企业版重磅升级：支持自定义AI Agent工作流
微软宣布Copilot Studio企业版重大更新，允许企业构建和部署自定义AI Agent工作流，与Microsoft 365、Azure服务深度集成。新增多Agent编排能力，支持跨系统自动化任务。已有超过50,000家企业部署自定义Copilot，月活用户突破1亿。
Notion AI Research Mode Combines Web Search with Company Documents
Notion launched AI Research Mode, enabling comprehensive research that simultaneously searches the web, company documents, and databases, synthesizing findings into structured documents with source attribution. The feature particularly popular with strategy, marketing, and product teams—early adopters report 60-70% reduction in time spent on background research. Research Mode integrates natively with Notion databases, with output able to automatically populate properties and trigger workflow automations. Available to Notion AI subscribers ($8/month).
AI Inference Costs Drop 90% Since 2023: Democratization Accelerates
A new analysis by Andreessen Horowitz documents the dramatic cost collapse in AI inference: GPT-4 class intelligence that cost $60 per million tokens in 2023 now costs under $3/million tokens, with continued decline expected. The drivers: NVIDIA Blackwell hardware efficiency gains, improved model architectures (mixture-of-experts, smaller context caches), quantization improvements, and intense provider competition. The practical implication: AI features that were economically viable only for high-value use cases in 2023 are now viable for nearly any software application. This is expected to trigger a second wave of AI product development targeting cost-sensitive use cases.
Perplexity AI Launches Enterprise Plan with Private Deployment
Perplexity AI launched Perplexity Enterprise with private model deployment, company knowledge base integration (Confluence, Notion, SharePoint), SSO, and admin controls—addressing the data privacy concerns that had held back corporate adoption. The enterprise tier targets research-intensive workflows: competitive intelligence, market research, and meeting preparation. Priced at $40/user/month, the launch is central to Perplexity's path to $50M ARR projected by end of 2025. The company recently raised additional funding at a valuation exceeding $1B.