Technical co-founder or outsourcing: how to decide技术联合创始人还是外包：如何决定

The decision is really about where your company's compounding advantage lives. If software is a cost center — an internal tool, a storefront, a integration layer — an agency or freelance team with a clear spec is the rational choice: you pay for output, the relationship ends when the deliverable ships, and nobody has to share equity. If software is the product, the iteration loop is the company: every week of customer learning has to turn into shipped changes, and renting that loop by the hour gets expensive in both money and speed.这个决定其实是在问：你公司的复利优势在哪里。如果软件是成本中心——一个内部工具、一个店面、一个集成层——那么一家有清晰规格的代理机构或自由职业团队就是理性选择：你为产出付费，交付物发出去关系就结束，没有人需要分股权。如果软件就是产品，迭代循环就是公司本身：每周从客户那里学到的东西都必须变成已上线的改动，按小时租这个循环，在金钱和速度上都会越来越贵。

The honest test is what happens after version one. Outsourcing builds what you specified; it does not sit in your customer calls, notice that the onboarding flow loses users at step three, and rebuild it over the weekend. A technical co-founder does — that is literally the job. So if your roadmap after launch is "keep changing everything based on what we learn," you need someone who owns that loop, not someone who invoices it. If your roadmap after launch is "it works, maintain it," outsourcing is fine and cheaper in equity terms.诚实的检验标准是第一版之后会发生什么。外包按你写好的规格来做；它不会坐进你的客户通话里，注意到引导流程在第三步流失用户，然后周末把它重做一遍。技术联合创始人会——这正是这份工作。所以如果上线后的路线图是「根据我们学到的不断改一切」，你需要一个拥有这个循环的人，而不是一个给这个循环开账单的人。如果上线后的路线图是「它能用，维护它」，外包就够了，从股权角度看也更便宜。

When the answer is co-founder, the failure mode is choosing on charm instead of evidence. Titles, decks, and enthusiasm are unverifiable; public work is not. GitHub proof-of-work — real commits, maintained repositories, shipped projects — is the strongest cheap signal that someone actually builds. This is what polyagent Builders (polyagent.club/builders) measures: it ranks builders who opted in by the public GitHub evidence behind a skill and names the repositories behind every result — so the "can this person actually build?" question is answered before you spend months finding out the hard way.当答案是联合创始人时，失败的方式是凭魅力而不是凭证据来选。头衔、演示文稿和热情都无法核实；公开的工作可以。GitHub 能力证明——真实的提交、持续维护的仓库、已上线的项目——是「这个人真的在做东西」最强、也最便宜的信号。这正是 polyagent Builders（polyagent.club/builders）衡量的：它按一项技能背后的公开 GitHub 证据，给主动加入的构建者排名，并列出每个结果背后的仓库——于是「这个人真的能做出来吗？」这个问题，在你花几个月用惨痛方式才发现之前，就已经有了答案。

A hybrid worth knowing: hire the agency for version one while you look for the co-founder. You de-risk the timeline and buy time to evaluate candidates on real collaboration — a small paid trial project together — instead of a leap of faith. The one path to avoid is outsourcing the core product indefinitely while equity terms age: the longer you wait, the more expensive the co-founder becomes, in every sense.有一种值得了解的混合做法：在你寻找联合创始人的同时，雇代理机构做第一版。你降低了时间线的风险，也换来时间，用真实的协作——一起做一个小的付费试做项目——来评估候选人，而不是凭信念一跃。要避开的一条路是无限期地把核心产品外包出去，同时股权条款在老化：你等得越久，联合创始人就变得越贵，无论从哪个意义上说都是如此。