slug: track-company-hiring
type: recipe
format: usecase
audience: research
status: published
title:
  en: See what a company is hiring for
  zh: 看一家公司在招什么人
dek:
  en: Read every open posting at one company and group it by role family, city and level, so a hiring plan shows up as numbers you can compare month to month.
  zh: 读一家公司所有在招的岗位，按岗位类别、城市和级别分组，把它的招聘计划变成能按月对比的数字。
datasets: [jobs]
integration: [rest, mcp]
topics: [ai-hiring, company-research]
regions: [US]
snapshot: 2026-09-30
published: 2026-09-30
exploration_credits: 57
agent_run: { low: 28, high: 35, cap: 50 }
highlights:
  - value: "123"
    label:
      en: open postings at Notion
      zh: 个 Notion 在招岗位
  - value: "41"
    label:
      en: of them are engineering roles
      zh: 个是工程类岗位
usecase:
  who:
    en: An investor preparing for a meeting, a competitor watching a rival, or a seller looking for a way in. A company's open roles say where it is putting money before any press release does.
    zh: 准备见面的投资人、盯着对手的竞品团队、想找切入点的销售。一家公司在招的岗位，比任何新闻稿都更早说明它把钱花在哪。
  figures:
    - file: families.json
      title:
        en: Notion's 123 open postings, by role family
        zh: Notion 的 123 个在招岗位，按岗位类别
      note:
        en: Families come from keywords in the title, first match wins, in the order listed in title-families.json. Each posting is counted once.
        zh: 类别按标题里的关键词划分，按 title-families.json 的顺序取第一个匹配。每个岗位只算一次。
      labels:
        data-ai: { en: "Data and AI", zh: "数据和 AI" }
        engineering: { en: "Engineering", zh: "工程" }
        product-design: { en: "Product and design", zh: "产品和设计" }
        sales: { en: "Sales and partnerships", zh: "销售和合作" }
        marketing: { en: "Marketing", zh: "市场" }
        customer: { en: "Customer and education", zh: "客户和培训" }
        g-and-a: { en: "Finance, legal and people", zh: "财务、法务和人力" }
        other: { en: "Other", zh: "其他" }
      legend:
        value: { en: "Open postings", zh: "在招岗位" }
    - file: cities.json
      title:
        en: Where the roles are
        zh: 岗位在哪
      note:
        en: The city on the posting. A remote posting counts where it is listed.
        zh: 岗位上写的城市。远程岗位按它挂的城市算。
      labels:
        San Francisco: { en: "San Francisco", zh: "旧金山" }
        New York: { en: "New York", zh: "纽约" }
      legend:
        value: { en: "Open postings", zh: "在招岗位" }
    - file: seniority.json
      title:
        en: At what level
        zh: 什么级别
      note:
        en: The seniority the posting states.
        zh: 岗位上写的级别。
      labels:
        Mid-Senior level: { en: "Mid to senior", zh: "中高级" }
        Entry level: { en: "Entry level", zh: "入门级" }
        Internship: { en: "Internship", zh: "实习" }
        Director: { en: "Director", zh: "总监" }
      legend:
        value: { en: "Open postings", zh: "在招岗位" }
  reads: [title, location.city, location.country, seniority, posted_date]
  next:
    - title: { en: "Run it again next month", zh: "下个月再跑一次" }
      body:
        en: The same prompt a month later shows which families grew. Keep each month's files; the difference is the signal.
        zh: 一个月后用同一段提示词再跑一次，就能看出哪些类别在增加。每个月的文件都留着，差值才是信号。
    - title: { en: "Compare two companies", zh: "两家公司对比" }
      body:
        en: Run it for a rival too. Two companies of similar size hiring very different mixes is worth a question in the meeting.
        zh: 对竞品也跑一次。规模相近的两家公司招人结构差很多，值得在会上问一句。
    - title: { en: "Read the postings that matter", zh: "细看关键岗位" }
      body:
        en: Ask the agent to summarize the descriptions in the family you care about, for example which tools the data roles ask for.
        zh: 让 agent 把你关心的那一类岗位的描述总结出来，比如数据岗位要求用哪些工具。
  faq:
    - q: { en: "Is this the company's headcount plan?", zh: "这就是这家公司的招聘计划吗？" }
      a:
        en: It is the open postings visible through the Metix AI Platform on one day. Some roles are never posted, and some postings stay up after they are filled.
        zh: 这是某一天通过 Metix AI Platform 能看到的在招岗位。有的岗位从来不公开招，有的招到人了也没下架。
    - q: { en: "Why read records instead of counting?", zh: "为什么要读记录而不只是计数？" }
      a:
        en: Role families come from the title text, which no filter groups for you. Reading 123 postings costs about 25 API Credits.
        zh: 岗位类别要看标题文字才能分，没有现成的筛选条件能直接分组。读 123 个岗位大约 25 API Credits。
    - q: { en: "Can I track a bigger company?", zh: "能看更大的公司吗？" }
      a:
        en: Yes, at 1 API Credit per 5 postings read. A company with 600 open postings costs about 125. The prompt stops and asks before its ceiling.
        zh: 可以，每读 5 个岗位 1 API Credit。600 个在招岗位的公司大约 125。提示词在上限之前会先停下来问你。
  method:
    - title: { en: "Population", zh: "统计范围" }
      body:
        en: Open postings with company.name "Notion" visible through the Metix AI Platform on September 30, 2026.
        zh: 2026 年 9 月 30 日通过 Metix AI Platform 可见、company.name 为 "Notion" 的在招岗位。
    - title: { en: "Grouping", zh: "分组" }
      body:
        en: Titles are matched against keyword lists in title-families.json in order, and each posting goes to its first match. Reposts are not collapsed.
        zh: 标题按 title-families.json 里的关键词顺序匹配，每个岗位归入第一个匹配的类别。重复发布的岗位不合并。
    - title: { en: "Limits", zh: "局限" }
      body:
        en: One day, not a trend. Keyword families are coarse; a "Solutions Engineer" counts as engineering here, not sales.
        zh: 只是一天，不是趋势。关键词分类比较粗；这里 "Solutions Engineer" 算工程，不算销售。