跳到主内容
@wquguru
精选88SaaStr 博客(RSS)产品与增长

Remote:AI接管薪酬与工程,人效增50%且零裁员

SaaStr AI App of the Week: Remote. The $300M Payroll Platform Your AI Agent Can Run

原文
发到 X
推荐理由

提供了 AI Agent 接入核心业务数据的 MCP 落地案例,以及“不裁员但停招”的人效提升具体路径,对独立开发者和 SaaS 团队极具参考价值。

Remote is a seven-year-old global payroll and employment platform based in Amsterdam. It leaned in early to making payroll (and more) for distributed and global teams work, and is one of the most trusted vendors in the space. Its payroll business has passed $300 million in ARR, growing more than 300% year over year, and it’s cash-flow positive. Use them. But it makes the AI App of the Week list for what it did with AI in 2026, not for the payroll business itself.

Remote 是一家总部位于阿姆斯特丹的全球薪酬与雇佣平台,已有七年历史。它较早致力于解决分布式和全球团队的薪酬(及更多)管理问题,是该领域最受信赖的供应商之一。其薪酬业务年度经常性收入(ARR)已突破 3 亿美元,同比增长超过 300%,且实现现金流为正。你可以使用它们的服务。但它入选本周 AI 应用榜单,是因为其在 2026 年对 AI 的应用,而非其薪酬业务本身。

The short version: Remote shipped an MCP server so AI agents can operate payroll directly, more than 85% of its code is now written by AI, and revenue per employee is up 50% without cutting a single job.

简而言之:Remote 发布了一个 MCP 服务器,使 AI 代理能够直接操作薪酬;目前超过 85% 的代码由 AI 编写,人均营收增长 50%,且未裁减任何岗位。

What’s new in 2026:

2026 年的新动态:

  • Remote MCP, an interface on the Model Context Protocol, gives AI agents secure access to payroll, contract, compliance and org data
  • Platforms like BambooHR and Workday can now use Remote as the underlying payroll engine
  • More than 85% of all code was written by AI in the last month, and engineering contributions are up 60%+ year over year
  • Revenue per employee is up 50% with no layoffs; hiring plans are being re-evaluated toward upskilling instead
  • Remote Build embeds forward-deployed engineers with customers to build AI workflows inside their orgs
  • Remote Labs, an internal AI marketplace, lets any employee ship tools on company infrastructure
  • AI-powered payroll now runs in more than 100 countries on proprietary calculation engines
  • Remote MCP 是基于模型上下文协议(Model Context Protocol)的接口,为 AI 代理提供对薪酬、合同、合规和组织数据的受控访问权限。
  • BambooHR 和 Workday 等平台现在可以将 Remote 作为底层薪酬引擎使用。
  • 在过去一个月中,超过 85% 的所有代码均由 AI 编写,工程贡献同比增幅超过 60%。
  • 在零裁员的情况下,人均营收增长 50%;招聘计划正重新评估,转向技能提升而非扩张。
  • Remote Build 将前置部署工程师嵌入客户内部,帮助其在组织内构建 AI 工作流。
  • Remote Labs 是一个内部 AI 市场,允许任何员工在公司基础设施上发布工具。
  • AI 驱动的薪酬服务现已在其专有计算引擎的支持下,覆盖全球 100 多个国家。

1. Remote MCP: your AI agent can run payroll

1. Remote MCP:你的 AI 代理可以运行薪酬流程

The headline feature is Remote MCP. Built on the Model Context Protocol, it lets AI agents and external platforms read and act on payroll and compliance data with access controls attached. BambooHR and Workday can plug Remote in as the engine underneath their own products.

核心功能是 Remote MCP。基于模型上下文协议构建,它允许 AI 代理和外部平台在附加访问控制的前提下,读取并操作薪酬和合规数据。BambooHR 和 Workday 可以将 Remote 集成到其自有产品之下作为底层引擎。

CEO Job van der Voort’s framing is blunt: connect ChatGPT or Claude and “you can control all of Remote,” to the point where “you don’t have to interact with our platform anymore.” He thinks that is where the future goes, and he is building for the version where the software disappears and the agent is the interface.

CEO Job van der Voort 的表述直截了当:连接 ChatGPT 或 Claude,你就可以“控制 Remote 的一切”,以至于“你不再需要与我们的平台进行交互”。他认为这是未来的方向,他正在为软件消失、代理成为界面的版本而构建。

Remote’s own team demos the practical version: the Chief People Officer, who describes herself as non-technical, built a working global case-management tracker with plain-English prompts, wired into live workforce and compliance data through Remote MCP. No engineering ticket, no implementation cycle:

Remote 自身团队展示了这一功能的实际应用版本:首席人力官自称非技术背景,通过自然语言提示词,结合 Remote MCP 接入实时劳动力数据和合规数据,构建了一个可运行的全球案例管理追踪器。无需工程工单,无需实施周期:

2. The CEO’s personal agent is the security model

2. CEO 的个人代理是安全模型

Payroll data is about as sensitive as corporate data gets, so the interesting part of Remote MCP is not the access, it is the containment. Van der Voort’s early explorer is his own open-source personal agent, which he named Jim.

薪资数据是敏感程度最高的企业数据之一,因此 Remote MCP 的有趣之处不在于访问权限,而在于隔离控制。Van der Voort 早期的探索者是他自己的开源个人智能体,他将其命名为 Jim。

His description of the setup: Jim can interact with Remote, “but he cannot do destructive things.” The agent gets the access it needs and nothing more. That scoped, non-destructive pattern is the whole product thesis. If agents are going to touch payroll, the permission model has to be built for agents that will, eventually, do something unexpected.

他对设置的描述如下:Jim 可以与 Remote 交互,“但它不能执行破坏性操作。”该智能体仅获取其所需的访问权限,不多不少。这种受限且非破坏性的模式是整个产品的核心论点。如果智能体要接触薪资数据,那么权限模型必须为那些最终可能会做出意外行为的智能体而构建。

3. More than 85% of the code is written by AI

3. 超过 85% 的代码由 AI 编写

Internally, Remote is one of the more aggressive adopters among companies its size. Engineering contributions are up more than 60% over the last year, and van der Voort says that in the last month, “more than 85% of all of our code is written by AI.” He describes the trend as accelerating, not plateauing.

在内部,Remote 是其规模同类公司中更积极采用 AI 的企业之一。工程领域的贡献在过去一年中增长了 60% 以上,van der Voort 表示,在上个月,“我们所有的代码中有超过 85% 是由 AI 编写的。”他将这一趋势描述为正在加速,而非趋于平稳。

Five Claude instances run on his second screen while he works, building things for him and for the company. The Slack agent that summarizes discussions started as one of these experiments. This is the part of the story that rhymes with every AI-native startup right now, except Remote is doing it at $300M+ in ARR with a compliance-heavy product where a bad deploy can mean someone doesn’t get paid.

他在工作的第二块屏幕上运行着五个 Claude 实例,让它们为自己和公司构建功能。用于总结讨论的 Slack 智能体最初就是这些实验之一。这是目前每个 AI 原生初创企业都在讲述的故事的一部分,但 Remote 是在 ARR(年度经常性收入)超过 3 亿美元、且合规要求极高的产品环境中这样做,而在这样的产品中,一次糟糕的部署可能导致有人收不到工资。

4. Revenue per employee is up 50%, with zero layoffs

4. 人均营收增长 50%,零裁员

The number van der Voort leads with is not ARR. It is revenue per employee, up 50% after AI adoption across every function, not just engineering. And unlike the Cloudflare story this week, where a 20% cut accompanied the efficiency gains, Remote says it has not cut jobs over AI.

van der Voort 首先提到的数字不是 ARR,而是人均营收。在 AI 被引入所有职能部门(而不仅仅是工程部门)后,人均营收增长了 50%。与本周 Cloudflare 的故事不同——后者在效率提升的同时裁减了 20% 的员工——Remote 表示它并未因 AI 而裁员。

What changed is hiring. Departments that planned to grow headcount are now asked a question first: “do we actually need more people,” or should the money go to upskilling the team and buying more AI instead? Some roles that would have been hired simply were not. The company frames it as deferred hiring rather than avoided hiring, which is honest about the direction of travel.

发生变化的是招聘计划。原本计划增加人员编制的部门现在首先要回答一个问题:“我们是否真的需要更多的人?”还是应该将资金用于团队技能提升和购买更多 AI?一些原本会招聘的职位 simply 没有被填补。公司将这种情况描述为“延迟招聘”而非“避免招聘”,这诚实地反映了发展方向。

Van der Voort’s own summary of his job as CEO now: make sure the company doesn’t run out of money and grows as fast as possible, with AI spend rising but tracked. “This adds a whole new fun angle, I would say.”

van der Voort 对自己作为 CEO 的工作的最新总结是:确保公司不会耗尽资金,并以尽可能快的速度增长,同时 AI 支出在上升但受到追踪。“我认为这增添了一个全新的有趣角度。”

5. Remote Build: forward-deployed AI engineers

5. Remote Build:前置部署的 AI 工程师

Remote knows it is ahead of most of its customers on internal AI adoption, so it productized the gap. Remote Build embeds engineers directly with customers and prospects to build AI-powered workflows inside their organizations. Van der Voort calls it what investors call it: forward-deployed engineers.

Remote 深知其在内部 AI 采用方面领先于大多数客户,因此将这一差距产品化。Remote Build 将工程师直接嵌入客户和潜在客户中,在其组织内部构建 AI 驱动的工作流程。Van der Voort 称其为投资者所称的:前置部署工程师。

The model will be familiar to anyone who watched Palantir. The difference is the domain: instead of data infrastructure, the engineers are wiring AI into payroll, compliance and HR operations, the exact processes most companies have never automated because the regulatory surface is terrifying. Remote Labs, the internal marketplace where employees ship tools on company infrastructure, is the pipeline that feeds it.

该模式对于任何看过 Palantir 的人来说都不会陌生。不同之处在于领域:工程师不是构建数据基础设施,而是将 AI 集成到薪酬、合规和人力资源运营中,这些正是大多数公司从未自动化的确切流程,因为监管范围令人望而生畏。Remote Labs(员工在公司基础设施上交付工具的内部市场)是为其提供支持的管道。

6. AI payroll is now live in more than 100 countries

6. AI 薪酬现已在 100 多个国家上线

The product underneath the AI story is still payroll, and Remote’s bet is that owning the hard parts outright is the moat. The company owns and operates 100% of its legal entities rather than routing through aggregators, and it built proprietary AI-powered calculation engines for payroll, now running in its most popular hiring countries and expanding.

AI 故事背后的产品仍然是薪酬,而 Remote 的赌注在于完全拥有最难的部分才是护城河。该公司拥有并运营其 100% 的法律实体,而不是通过聚合商中转,并且它构建了专有的 AI 驱动计算引擎用于薪酬处理,目前已在最受欢迎的招聘国家运行并正在扩展。

The pitch against the old model: multi-country payroll through aggregators means delays, errors and compliance risk from chains of local vendors. Remote’s version processes instantly with the compliance logic in the engine. Newer additions include AI-powered employee surveys that run engagement analysis in under a minute and AI-supported recruiting.

针对旧模式的论点:通过聚合商进行多国薪酬意味着延迟、错误以及由本地供应商链条带来的合规风险。Remote 的版本通过引擎中的合规逻辑即时处理。新增功能包括 AI 驱动的员工调查,可在不到一分钟的时间内完成参与度分析,以及 AI 支持的招聘。

Van der Voort’s line on scope: “We do payroll for everybody, period.” Not just distributed teams. Most of Remote’s customers have people in offices.

Van der Voort 关于范围的表述:“我们为所有人提供薪酬服务,仅此而已。”不仅仅是分布式团队。Remote 的大多数客户在办公室都有人员。

Zapier, the AI orchestration platform, is one of them. It hired into 8 countries through Remote’s EOR in the last six months, cut about 80% of the time spent on manual payroll data entry after integrating BambooHR with Remote, and gets compliance guidance in under 30 minutes instead of several hours:

Zapier(AI 编排平台)就是其中之一。在过去六个月里,它通过 Remote 的 EOR 在 8 个国家进行了招聘,在与 BambooHR 集成后,手动薪酬数据输入所花费的时间减少了约 80%,并且获取合规指导的时间从数小时缩短到不到 30 分钟:

7. Where it fits and where it doesn’t

7. 适用与不适用的场景

Remote fits companies hiring across borders that want payroll to disappear into infrastructure, and now, teams that want agents to operate that infrastructure through MCP rather than clicking through a dashboard.

Remote 适合希望薪酬融入基础设施的跨国招聘公司,以及现在希望通过 MCP 而非点击仪表板来操作该基础设施的团队。

MCP-based access to payroll is new, and the security model, however well scoped, is unproven at enterprise scale with auditors watching. Arguably, start agents read-only. Second, the category is crowded: Deel, Rippling and a dozen others all claim the AI payroll story now, and Remote’s answer is depth of owned infrastructure, which is real but slower to evaluate than a feature checklist.

基于 MCP 访问薪酬数据是全新的做法,尽管安全模型的范围界定良好,但在审计人员密切关注的企业级规模下尚未经过验证。可以说,应让智能体以只读模式启动。其次,该领域竞争者众多:Deel、Rippling 和十余家其他公司都声称拥有 AI 薪酬解决方案,而 Remote 的回应在于其自有基础设施的深度,这确实存在,但比功能清单更难评估。

8. The “disappear in a good way” thesis

8. “以良好方式消失”的理论

As agentic AI rises, many companies could “virtually disappear, in a good way.” Remote is unusual in saying this about itself on the record. Most vendors pitch the agent as a feature of their platform. Remote is pitching the platform as a backend the agent calls, with BambooHR and Workday as the first evidence.

随着智能体 AI 的兴起,许多公司可能会“以良好方式虚拟消失”。Remote 罕见地在公开场合对此进行自我陈述。大多数供应商将智能体作为其平台的一项功能进行推广。Remote 则将平台定位为智能体调用的后端,BambooHR 和 Workday 作为首批证据。

更进一步:量化金融体系

看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力

进入量化体系 →

相似阅读

关联信息,但可能不是同一事件