Software companies are changing. They are using new AI tools and buying other companies.

軟體公司正在改變。他們正利用新的 AI 工具並收購其他公司。

Salesforce is buying AI companies. They paid 3.6 billion dollars for a company called Fin. Some people think this is good. Other people think the company will lose money.

Salesforce 正在收購 AI 公司。他們花了 36 億美元收購了一家名為 Fin 的公司。有些人認為這是好事，而有些人則認為公司會虧損。

Some new companies are very small. One company has only ten workers. They use AI to write code. They made 13 million dollars. But the workers forget how to code by hand.

一些新公司規模非常小。有一家公司僅有十名員工，他們使用 AI 撰寫程式碼，創造了 1,300 萬美元的收入。但員工們卻逐漸忘記如何手寫程式碼。

Another company is called Arena. It helps people test AI. This company grew very fast. It made 100 million dollars in eight months.

另一家名為 Arena 的公司則協助人們測試 AI。這家公司的成長速度非常快，在八個月內就創造了 1 億美元的收入。

Old companies are trying to change. New AI companies are growing fast.

老牌公司正嘗試轉型，而新的 AI 公司則成長迅速。

The software industry is currently moving toward 'agentic' artificial intelligence. This change is marked by large companies buying smaller firms and the rise of small, AI-focused startups.

軟體產業目前正朝向「代理型」人工智慧發展。這一轉變的標誌是大公司收購小公司，以及專注於 AI 的小型初創企業崛起。

Salesforce has started a series of strategic acquisitions to improve its AI tools, most notably the $3.6 billion purchase of Fin, an AI customer service platform. Along with smaller deals for M3ter and Contentful, these moves aim to strengthen the Agentforce suite. However, some market experts are skeptical. For example, analyst Gil Luria emphasized that AI might disrupt the traditional way companies make money from software licenses, which may explain why Salesforce's stock fell by about 40% last year. On the other hand, Cantor Fitzgerald asserted that established software providers could still dominate the market by buying innovative AI startups that lack the size to grow on their own.

Salesforce 已開始一系列策略性收購以提升其 AI 工具，最著名的是以 36 億美元購入 AI 客服平台 Fin。加上對 M3ter 和 Contentful 的小型交易，這些舉措旨在強化 Agentforce 套件。然而，部分市場專家對此持懷疑態度。例如，分析師 Gil Luria 強調，AI 可能會顛覆公司從軟體授權獲利的傳統方式，這或許解釋了為何 Salesforce 的股價去年下跌了約 40%。另一方面，Cantor Fitzgerald 則斷言，成熟的軟體供應商只要收購那些規模不足以獨立成長的創新 AI 初創公司，仍能主導市場。

At the same time, a new trend of 'tiny teams' has appeared. One example is an AI note-taking app started by students that earned $13 million in total revenue with only ten full-time employees. They used AI coding tools to stay highly productive, although the founders warned that relying too much on automation could weaken basic coding skills. This shows a broader shift where AI allows very small teams to produce work that previously required many more people.

與此同時，出現了一種「微型團隊」的新趨勢。其中一個例子是由學生創立的 AI 筆記應用程式，僅用十名全職員工就創造了 1,300 萬美元的總營收。他們利用 AI 程式碼工具來保持高生產力，儘管創辦人警告，過度依賴自動化可能會削弱基礎的程式碼撰寫能力。這顯示出一個更廣泛的轉移，即 AI 允許極小規模的團隊產出過去需要更多人力才能完成的工作。

Furthermore, the business of evaluating AI is growing quickly. Arena, a company from UC Berkeley, reached $100 million in annual revenue just eight months after launching. By selling data on how AI models perform, Arena is now competing with other firms like Scale AI. This growth shows that there is a high demand for services that help AI providers refine and optimize their models for complex tasks.

此外，AI 評估業務正快速成長。來自加州大學柏克萊分校的 Arena 公司在推出僅八個月後，年營收便達到 1 億美元。透過銷售 AI 模型表現的數據，Arena 目前正與 Scale AI 等公司競爭。這種增長顯示出市場對於協助 AI 供應商精鍊並優化模型以處理複雜任務的服務有高度需求。

The software sector is currently split between older companies trying to change through acquisitions and flexible, AI-native startups that are growing rapidly.

軟體產業目前分為兩派：一方是試圖透過收購進行轉型的老牌公司，另一方則是快速成長且靈活的 AI 原生初創公司。

The software industry is currently experiencing a transition toward agentic artificial intelligence, characterized by aggressive corporate acquisitions and the emergence of lean, AI-native startups.

軟體產業目前正經歷向代理型人工智慧（agentic AI）的轉型，其特徵為積極的企業收購以及精簡的 AI 原生初創公司的興起。

Salesforce has initiated a series of strategic acquisitions to enhance its artificial intelligence capabilities, most notably the $3.6 billion procurement of Fin, an AI customer service platform. This transaction, alongside smaller acquisitions of M3ter and Contentful, is intended to augment the Agentforce suite. Despite these efforts, the organization faces significant market skepticism. Analysts, such as Gil Luria, suggest that the shift toward AI may disrupt the traditional seat-based revenue model, a sentiment reflected in the company's stock performance, which declined approximately 40% over the past year. Conversely, Cantor Fitzgerald posits that established SaaS providers may achieve market dominance by acquiring innovative, though smaller, AI-native firms that lack necessary scale.

Salesforce 已啟動一系列策略性收購以強化其人工智慧能力，最顯著的是以 36 億美元收購 AI 客戶服務平台 Fin。此次交易以及對 M3ter 和 Contentful 的小型收購，旨在增強 Agentforce 套件。儘管如此，該組織仍面臨巨大的市場質疑。分析師 Gil Luria 等人認為，向 AI 的轉型可能會顛覆傳統的按席位計費收入模式，該公司過去一年股價下跌約 40% 亦反映了此種情緒。相反地，Cantor Fitzgerald 認為，成熟的 SaaS 提供商可以透過收購缺乏必要規模但具創新力的 AI 原生小公司來實現市場主導地位。

Parallel to these institutional shifts, a new paradigm of 'tiny teams' has emerged, exemplified by a student-founded AI note-taking application. This venture achieved $13 million in lifetime revenue with only ten full-time employees, utilizing AI code-generation tools to maintain high productivity. However, the founders noted a potential atrophy of fundamental coding skills resulting from an over-reliance on automated systems. This trend underscores a broader shift where AI enables minimal human staffing to produce output previously requiring significantly larger teams.

與這些機構轉變平行，一種「微型團隊」的新典範已經出現，例如由學生創辦的 AI 筆記應用程式。該創業公司僅憑 10 名全職員工，利用 AI 代碼生成工具維持高生產力，實現了 1,300 萬美元的終身收入。然而，創辦人指出，過度依賴自動化系統可能會導致基礎編碼技能的萎縮。這一趨勢強調了一個更廣泛的轉變，即 AI 使得極少數的人力配置即可產生先前需要龐大團隊才能完成的產出。

Furthermore, the commercialization of AI evaluation is accelerating. Arena, a UC Berkeley spin-off, reached a $100 million annualized run-rate revenue within eight months of launching its commercial services. By monetizing crowdsourced model performance analytics, Arena competes for capital with human-labeling firms such as Scale AI and Mercor. This growth reflects an increasing institutional demand for post-training refinement services as AI providers seek to optimize model performance across complex workflows.

此外，AI 評估的商業化正在加速。加州大學柏克萊分校（UC Berkeley）的衍生公司 Arena 在推出商業服務後的八個月內，年化運行收入便達到 1 億美元。透過將群眾外包的模型性能分析貨幣化，Arena 與 Scale AI 和 Mercor 等人力標記公司競爭資本。這種增長反映出，隨著 AI 提供商尋求優化複雜工作流中的模型性能，機構對訓練後精煉服務的需求日益增加。

The sector remains divided between legacy providers attempting to pivot through M&A and agile, AI-native entities achieving rapid scalability.

該產業目前分裂為兩派：一派是嘗試透過併購進行轉型的傳統供應商，另一派則是能實現快速擴展的靈活 AI 原生實體。