AI for the Fortune 2000: How Tenex Builds an AI-Native SDLC — Alex Lieberman & CJ Hess面向财富 2000 强的 AI：Tenex 如何搭建 AI 原生的软件开发流程

面向财富 2000 强的 AI：Tenex 如何搭建 AI 原生的软件开发流程AI for the Fortune 2000: How Tenex Builds an AI-Native SDLC — Alex Lieberman & CJ Hess

Key points

• Tenex pays engineers by story points rather than hours; it has grown to 75 people, with half its customers in the Fortune 2000.
  Tenex 按故事点而非工时给工程师付酬，已发展到 75 人，一半客户是财富 2000 强。

• Its MetaHarness checks plans, decisions and evidence into Git alongside code, so agents can read why work was done, not just what.
  其 MetaHarness 把计划、决策和证据与代码一起提交到 Git，让 Agent 不仅知道做了什么，还知道为什么。

• Enterprise adoption funnel: ~90% have rolled out Claude or Codex, ~50% coding agents, but only ~5% have agents in production for non-engineering work.
  企业落地漏斗：约 90% 已部署 Claude 或 Codex，约 50% 用上编程 Agent，但只有约 5% 在非工程工作中有生产级 Agent。

• Alex is broadly bearish on vertical AI products whose main feature is an agent doing a task; swyx pushes back with Harvey and OpenEvidence.
  Alex 总体看空以“Agent 完成某任务”为主卖点的垂直 AI 产品；swyx 以 Harvey、OpenEvidence 为例反驳。

Quotes

• “The thesis is that AI is really just a leveraged bet on a human’s intellect and work ethic.”
  “我们的论点是：AI 其实只是对一个人的智识和职业态度的杠杆押注。”

• “You can’t automate a process that does not exist.”
  “你无法自动化一个并不存在的流程。”