<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[SV Vector 2026-03]]></title><description><![CDATA[SV Vector 2026-03]]></description><link>https://www.sv-vector.com/insights</link><generator>RSS for Node</generator><lastBuildDate>Wed, 03 Jun 2026 16:10:17 GMT</lastBuildDate><atom:link href="https://www.sv-vector.com/en/blog-feed.xml" rel="self" type="application/rss+xml"/><item><title><![CDATA[Silicon Valley Vector x Yuandong Tian, Deep Dive on AI]]></title><description><![CDATA[“The flood is coming, yet many are still living in a 'quiet years' illusion.” —— Yuandong Tian, Former Research Director at Meta AI On February 28, 2026, Qinyun Cao, host of Silicon Valley Coordinates, interviewed Yuandong Tian in Silicon Valley. Having spent 11 years at Meta, Tian led pioneering research in reinforcement learning, LLM reasoning, and long context. He recently embarked on a new entrepreneurial journey as a Co-founder. This conversation covers LLM competition, memory and...]]></description><link>https://www.sv-vector.com/en/post/silicon-valley-vector-x-yuandong-tian-deep-dive-on-ai</link><guid isPermaLink="false">69aa4c681579f653138bcded</guid><pubDate>Fri, 06 Mar 2026 04:53:53 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/e9a992_5c2f90768d444ae7a191c98aaee9d360~mv2.png/v1/fit/w_1000,h_1000,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>Silicon Valley Vector</dc:creator></item><item><title><![CDATA[硅谷坐标 x 田渊栋深度访谈]]></title><description><![CDATA[“洪水马上来了，很多人还岁月静好。” —— 田渊栋，前Meta AI研究总监 2026年2月28日，硅谷坐标主持人曹卿云在硅谷专访了田渊栋。田渊栋在Meta深耕11年，主导了强化学习、大模型推理与长上下文等多项前沿研究，近期刚刚以Co-founder身份开启新的创业旅程。本次对话涵盖大模型竞争、记忆与存储、推理前沿路线、Agents对社会的冲击等议题，以下是核心内容整理。 田渊栋 CMU机器人学博士。在Meta AI（FAIR）深耕11年，担任研究总监。他主导了开源围棋AI ELF OpenGo（单GPU击败职业棋手）、大模型长上下文扩展的奠基工作 Positional Interpolation 与 Attention Sinks（ICLR 2024，引用逾1,400次），以及隐空间推理前沿探索 Coconut，同期联席主导 Llama 4 Reasoning 方向，Google Scholar总引用超22,000次，h-index 74。近期以Co-founder身份开启新的创业旅程。 「本期核心观点」 1. 大模型竞争：蒸馏时代，领先优势正在缩短...]]></description><link>https://www.sv-vector.com/post/%E7%A1%85%E8%B0%B7%E5%9D%90%E6%A0%87-x-%E7%94%B0%E6%B8%8A%E6%A0%8B%E6%B7%B1%E5%BA%A6%E8%AE%BF%E8%B0%88</link><guid isPermaLink="false">69aa49e4b5dd058733480b4a</guid><pubDate>Fri, 06 Mar 2026 04:52:14 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/e9a992_5c2f90768d444ae7a191c98aaee9d360~mv2.png/v1/fit/w_1000,h_1000,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>Silicon Valley Vector</dc:creator></item><item><title><![CDATA[The SaaS Re-valuation: AI Upends the Growth Narrative]]></title><description><![CDATA[Over the past year, many SaaS companies have experienced a counter-intuitive phenomenon: while retention metrics (GRR/NRR) don’t appear to be collapsing, the market is still pricing "future growth" at significantly lower multiples. This feels less like a market crash and more like a discount on growth certainty. Since the emergence of AI, investors have arrived at a very different set of answers regarding how software will continue to scale in the future. "Workflow Differentiation / Risk...]]></description><link>https://www.sv-vector.com/en/post/the-saas-re-valuation-ai-upends-the-growth-narrative</link><guid isPermaLink="false">69a6956804e15cb50901ffe6</guid><pubDate>Tue, 03 Mar 2026 08:04:38 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/4605ed_df270c9e9ace4b9e9e0099454ffcc500~mv2.jpg/v1/fit/w_897,h_897,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>JJ S</dc:creator></item><item><title><![CDATA[The Rise of Vertical Agents: Basis Hits Unicorn Status with $100M Raise for Accounting AI]]></title><description><![CDATA[Another noteworthy signal comes from the vertical sector: Basis , an AI agent platform tailored for accounting firms, has announced a $100M funding round at a $1.15B valuation  to accelerate product development and team expansion. The platform focuses on using agents to automate structured accounting workflows—including tax, audit, and advisory services—by integrating Large Language Models (LLMs) with rule-based controls, audit trails, and compliance frameworks, all while maintaining...]]></description><link>https://www.sv-vector.com/en/post/the-rise-of-vertical-agents-basis-hits-unicorn-status-with-100m-raise-for-accounting-ai</link><guid isPermaLink="false">69a694dab09ae9dfda0be474</guid><pubDate>Tue, 03 Mar 2026 08:01:18 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/4605ed_076e22fabc1d40fc8011aecea4ec0f06~mv2.jpg/v1/fit/w_1000,h_1000,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>JJ S</dc:creator></item><item><title><![CDATA[垂直 Agent 正在加速：会计 AI 平台 Basis 融资 $100M、估值 $1.15B]]></title><description><![CDATA[一个值得关注的信号来自垂直场景：面向会计师事务所的 agent 平台 Basis 宣布融资 $100M，估值 $1.15B，用来加速产品与扩张团队。它主打用 agent 自动化结构化会计工作流（税务、审计、咨询等），把 LLM 与规则控制、审计追踪、合规框架结合起来，并保持 human-in-the-loop 的审批与复核。文章还提到其客户中包含约 Top 25 会计师事务所的约 30%。  SVV 视角：这条融资更像“路线图被验证” 不是所有 SaaS 都同等被 AI 冲击；相反，强流程、强合规、强审计链的领域，反而更适合做“端到端 agent + 人类复核”的落地形态。会计/税务/审计正是典型：任务结构化、错误成本高、但可通过流程约束与审计追踪来降风险。 这类产品的护城河往往不是“模型本身”，而是：工作流编排 + 合规规则 + 数据接入/数据结构 + 可追溯产物（audit trail），以及与事务所现有系统的集成深度。]]></description><link>https://www.sv-vector.com/post/%E5%9E%82%E7%9B%B4-agent-%E6%AD%A3%E5%9C%A8%E5%8A%A0%E9%80%9F%EF%BC%9A%E4%BC%9A%E8%AE%A1-ai-%E5%B9%B3%E5%8F%B0-basis-%E8%9E%8D%E8%B5%84-100m%E3%80%81%E4%BC%B0%E5%80%BC-1-15b</link><guid isPermaLink="false">69a6574d7db715abddf84b32</guid><pubDate>Tue, 03 Mar 2026 03:36:59 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/4605ed_076e22fabc1d40fc8011aecea4ec0f06~mv2.jpg/v1/fit/w_1000,h_1000,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>JJ S</dc:creator></item><item><title><![CDATA[SaaS 被“重新估值”：AI 颠覆增长叙事]]></title><description><![CDATA[过去一年，很多 SaaS 公司出现了一个反直觉现象：留存（GRR/NRR）看起来并没有崩，但市场仍在用更低的倍数定价“未来增长”。这更像是一次增长确定性的折价——AI 出现后，投资人对“软件未来如何继续增长”有了不同答案。更关键的是，买方预算正在从“再叠一个系统/模块”转向“AI 工具 + AI 改造”，因为后者的 ROI 叙事更直接：更少的人做更多的事，效率收益更快可见。短期结果是：不少 ISV 的增速被注意力迁移压住；中长期悬念变成：谁能把 AI 变成可持续收入，谁又会被 agent 贴着 API 把价值层绕走。 “工作流分化/风险象限”类框架图（示意图） SVV 的采访启示：给经营者的“本周两条硬要求” 产品必须 AI 化：不是“加一个 AI 功能”，而是能说清楚 哪些工作从人→机器、价值如何计量、客户为何愿意付费。并且要能落到可销售的 package（功能、定价、交付、风控与责任边界）。 运营也必须 AI 化：把 AI 用在内部“提速”上——更快决策、更快交付、更快试错；否则外部范式在变、内部时钟还停在旧节奏，差距会指数级拉开。 一个董事会级别的自检问题（12–18...]]></description><link>https://www.sv-vector.com/post/saas-%E8%A2%AB-%E9%87%8D%E6%96%B0%E4%BC%B0%E5%80%BC-%EF%BC%9Aai-%E9%A2%A0%E8%A6%86%E5%A2%9E%E9%95%BF%E5%8F%99%E4%BA%8B</link><guid isPermaLink="false">69a63245fe2730c04135613f</guid><pubDate>Tue, 03 Mar 2026 00:59:50 GMT</pubDate><enclosure url="https://static.wixstatic.com/media/4605ed_df270c9e9ace4b9e9e0099454ffcc500~mv2.jpg/v1/fit/w_897,h_897,al_c,q_80/file.png" length="0" type="image/png"/><dc:creator>JJ S</dc:creator></item></channel></rss>