How to Build an AI-Native Services Company

"Some of the biggest companies of the next decade won't be software businesses at all."

"下一个十年,一些最大的公司根本不会是软件公司。"

"They'll be services companies like insurance carriers and law firms rebuilt from scratch with AI doing most of the work."

"它们会是服务公司——保险公司、律师事务所这类——从零重建,大部分活由 AI 来干。"

"These are what we call AI native service companies. And the markets are trillions of dollars in size."

"这就是我们说的 AI 原生服务公司(AI native service companies)。而这些市场的规模是数万亿美元。"

"Tax, audit, insurance, law, parts of health care, and so forth."

"税务、审计、保险、法律、医疗的一部分,等等。"

"This opportunity didn't exist even a couple years ago."

"这个机会哪怕在两年前都还不存在。"

"But advances in the models have unlocked this new type of business where companies provide the outcome to the customer versus build a co-pilot that the customer uses internally."

"但模型的进步解锁了这种新型生意:公司直接把结果交付给客户,而不是造一个 co-pilot 让客户自己在内部用。"

"These companies also look and feel different than most startups today."

"这类公司在气质和形态上,也跟今天大多数创业公司不一样。"

"In this video, I'll walk through a playbook for founders starting AI services businesses from scratch."

"这个视频里,我会带大家过一遍从零做 AI 服务生意的打法(playbook)。"

"It's aimed at people thinking about starting a company, not if you're already running one."

"它面向的是正在考虑创业的人,而不是已经在经营公司的人。"

"I'll share some obvious and non-obvious elements of building these businesses that we've observed here at YC."

"我会分享一些我们在 YC 观察到的、做这类生意时有的显而易见、有的没那么显而易见的要点。"

"Topics include picking a market, forming a team, building the actual product, serving the customers, the P&L, and whether or not you should even buy a business."

"话题包括:选市场、组建团队、真正把产品做出来、服务客户、利润表(P&L),以及你到底该不该去收购一家现成的公司。"

"One general comment before we get started. We're still early here."

"开始前先说一句总体的话:我们现在还处在很早的阶段。"

"Like most things in AI, the market is moving fast. We're learning as we go, but the early successes here should get you really excited."

"跟 AI 里的大多数东西一样,这个市场变化很快。我们也是边做边学,但这里的早期成功案例,应该足以让你非常兴奋。"

"First, picking the right market. The same general advice for all startups applies here with some important caveats."

"第一,选对市场。给所有创业公司的通用建议在这里同样适用,但有几个重要的前提要注意。"

"You should pick a market you're excited to work in for a long time. These companies still take a decade or more."

"你要挑一个自己愿意长期投入、会为之兴奋的市场。这类公司照样要花十年甚至更久。"

"If you don't love some combination of the customers or the market or the technical problem, you're not going to make it."

"如果你对客户、市场、技术难题这几样里的某种组合没有真正的热爱,你是撑不下去的。"

"And that part isn't really new, but the best markets for AI services have four new, pretty unique traits."

"这一点其实不新鲜。但最适合做 AI 服务的市场,有四个全新的、相当独特的特征。"

"The first is low trust, meaning the work is already outsourced and the customer cares about the final product, not how they got there."

"第一个是低信任(low trust)——意思是这活儿本来就已经外包出去了,客户只关心最终产出,不关心过程是怎么来的。"

"You're displacing a vendor, not asking the customer to do something fundamentally different."

"你是在替换掉一个供应商,而不是要求客户去做一件根本不同的事。"

"That's a huge deal because you're not changing behavior. You're showing up where the budget already lives and doing the work."

"这一点非常关键,因为你不需要改变客户的行为。你只是出现在预算本来就在的地方,把活干了。"

"Second, low judgment at the task level. If you can break the work into pieces and every piece needs a human exercising actual judgment, you can't really scale."

"第二,任务层面的低判断(low judgment)。如果你把工作拆成一块块,而每一块都需要一个人来做真正的判断,那你根本没法规模化。"

"You need most of the steps to be automatable with judgment focused in a few places where humans stay in the loop."

"你需要大部分步骤都能自动化,把需要判断的环节收拢到少数几处,让人留在这几处的闭环里(human in the loop)。"

"The third is a high intelligence threshold. This sounds contradictory, but actually it isn't."

"第三,高智能门槛(high intelligence threshold)。这听起来像是自相矛盾,但其实不是。"

"The overall work has to be hard, hard enough that models plus humans are needed to actually deliver an outcome the customer accepts."

"整体这活儿必须够难——难到需要"模型 + 人"配合,才能真正交付一个客户愿意接受的结果。"

"The fourth is regulation can actually be good. Regulated industries have higher expectations and legal accountability that raises the bar and the moat for founders."

"第四,监管其实可以是好事。受监管的行业有更高的预期和法律问责,这反而抬高了门槛,也抬高了创始人的护城河(moat)。"

"For instance, Panacea is a current YC company that provides FDA regulatory services for biotechs and medtechs."

"举个例子,Panacea 是一家现在在 YC 的公司,为生物科技和医疗科技公司提供 FDA 合规服务。"

"They actually hire experienced FDA consultants, pair them with an AI platform to deliver faster, higher quality FDA approvals."

"他们真的会去雇有经验的 FDA 顾问,把他们和一个 AI 平台搭在一起,更快、更高质量地拿下 FDA 审批。"

"So, what are some of the specific markets we have in mind here at YC? The known good fit markets include tax, audit, insurance, mortgages, parts of healthcare, and parts of logistics."

"那么,我们在 YC 具体看好哪些市场?已知契合度不错的市场包括:税务、审计、保险、抵押贷款、医疗的一部分,以及物流的一部分。"

"But there are plenty more markets nobody has touched yet. Don't hold yourself to the obvious ones or what people talk about on X."

"但还有大量没人碰过的市场。别把自己局限在那些显而易见的、或者大家在 X 上讨论的市场里。"

"And here are a few more things to keep an eye on when it comes to the markets. The first is on the models. Will the models disrupt these businesses?"

"关于市场,还有几件事要盯着。第一件是关于模型:模型会不会反过来颠覆这些生意?"

"It depends on what I call the Sam Altman test. You should ask yourself, as the models get better, does your service get stronger or does the model itself commoditize you?"

"这取决于我称之为"Sam Altman 测试"的东西。你应该问自己:随着模型变强,你的服务是跟着变强,还是模型本身把你商品化(commoditize)了?"

"You want to be in the first camp."

"你想待在第一种阵营里。"

"Where to be careful. Anything involving equipment and onsite labor. The software margin math doesn't apply when you own and operate physical things."

"哪里要小心:任何牵扯到设备和现场人力的东西。当你自己拥有并运营实物资产时,软件那套利润率的算法就不成立了。"

"It's very hard to create real leverage, though these can be really good businesses. let's leave this area to the robotics founders."

"要造出真正的杠杆(leverage)非常难——尽管这些也可以是很好的生意。这块地盘,还是留给做机器人的创始人吧。"

"One more honesty check. Ask yourself sincerely, are you using humans cuz the work genuinely needs judgment or you compensating for product gaps?"

"再做一个诚实的自检。真心问问自己:你用人,是因为这活儿真的需要判断,还是在拿人来弥补产品的短板?"

"Be honest here so you're not papering over product shortcomings with actual humans."

"这里一定要诚实,别用真人去糊住产品的缺陷。"

"There are still great massive technology businesses to be built with humans in the loop."

"当然,人留在闭环里,照样能做出伟大的、体量巨大的科技公司。"

"Second, and maybe most importantly, the right founding team. The same advice applies for all startups here, but again with some important caveats."

"第二,也许是最重要的一点:对的创始团队。跟所有创业公司的建议一样,但同样有几个重要前提。"

"You should build companies with people you already know and you've worked with."

"你应该和你已经认识、已经共事过的人一起创业。"

"If you're solo, think about the best people you've ever worked with and ask them to join you. You'd be surprised who says yes."

"如果你是单枪匹马,那就想想你共事过的最厉害的那些人,去邀请他们加入。谁会答应,可能会让你意外。"

"For AI services specifically, there's three attributes that all the best founders share. The first is domain fluency."

"具体到 AI 服务,最好的创始人都有三个共同特质。第一个是行业熟稔度(domain fluency)。"

"Direct experience is best, but learned is actually okay. You're selling to skeptical buyers and often regulated spaces."

"亲身经历当然最好,但后天学来的其实也行。你面对的是持怀疑态度的买家,而且往往是受监管的领域。"

"You have to bleed credibility. How you acquire it matters somewhat less."

"你必须浑身透着可信度、张口就是行话。至于这份可信度是怎么攒来的,反而没那么重要。"

"The second is model fluency. You need to know what frontier models can do today and design the product to ride the curve as they get better."

"第二个是模型熟稔度(model fluency)。你得清楚前沿模型今天能做到什么,并把产品设计成能随着模型变强而一起顺着曲线往上走。"

"There is no substitute for great tech here. People underestimate this."

"在这件事上,过硬的技术无可替代。很多人低估了这一点。"

"Next is operational rigor. Topics like variance, throughput, cycle times, SOPs, this is not an exciting set of words for most founders, but you are fundamentally running an operation."

"第三个是运营的严谨性(operational rigor)。波动、吞吐量、周期时间、SOP 这些词,对大多数创始人来说毫无吸引力,但你本质上就是在经营一套运营体系。"

"You have to learn that skill set and you have to enjoy it or at least you have to respect it. The product is an operation."

"你必须学会这套本事,而且要享受它——至少也得敬重它。产品本身就是一套运营。"

"A great example here is the general legal team, which is an AI native law firm that YC recently backed."

"一个很好的例子是 General Legal,这是一家 AI 原生律所,YC 最近投了它。"

"The founders have a unique mix of actual law firm experience at Cooley and Fenwick as well as years of technical leadership CaseText."

"几位创始人有一种独特的组合:既有在 Cooley 和 Fenwick 这些律所的实战经验,又有在 CaseText 多年的技术带队经历。"

"But most importantly, they think deeply about throughput and how they staff their firm."

"但最重要的是,他们会深入地琢磨吞吐量,以及怎么给律所配置人手。"

"They've integrated shift work into how they serve clients to reduce cycle times and attract the best lawyers on the team. This is a win-win for scale."

"他们把轮班制融进了服务客户的方式里,以缩短周期时间,并吸引最优秀的律师加入团队。对规模化来说这是双赢。"

"Now, let's talk about building the actual product. With AI services, the setup is the opposite of most software."

"现在来聊聊怎么真正把产品做出来。做 AI 服务,整个格局跟大多数软件正好相反。"

"The human is the interface of the customer, not the product. The product helps the human scale their work non-linearly."

"面向客户的界面是人,不是产品。产品的作用,是帮这个人把工作非线性地放大。"

"That changes pretty much everything around building the actual product."

"这几乎改变了做产品这件事的方方面面。"

"First, you need to apply an operations mindset. Find the bottlenecks and build for the bottlenecks."

"第一,你要用一种运营的思维方式。找到瓶颈,然后针对瓶颈来造产品。"

"Throughput and cycle time are now product metrics. Track them like you would daily active users."

"吞吐量和周期时间,现在成了产品指标。要像盯日活(DAU)一样去盯它们。"

"Variance is the existential problem here. By variance, I mean non-uniform outputs from your actual service."

"波动(variance)才是这里生死攸关的问题。我说的波动,是指你实际服务产出的不一致、不均匀。"

"Customers will fire you for variance faster than they will fire you for being a bit slower or a bit more expensive than the incumbents."

"客户会因为你结果不稳定而炒掉你,比因为你比在位者稍慢一点、稍贵一点而炒掉你要快得多。"

"They need to trust the output. Inconsistency destroys trust, which causes churn."

"他们需要信任你的产出。不一致会摧毁信任,而信任一没,就带来流失(churn)。"

"Thirdly, humans in the loop should scale non-linearly. If revenue scales just in line with the number of humans you add, you'll have major problems."

"第三,闭环里的人应该是非线性扩张的。如果你的收入只是随着你加人的数量同比例增长,那你会有大麻烦。"

"The humans in the loop also need to enjoy the software. They are your users."

"闭环里的这些人也得喜欢用你的软件——他们就是你的用户。"

"A general point, it's okay to do things that don't scale at the very beginning. But eventually, you really do need to scale."

"一个通用的点:一开始做些无法规模化的事没关系。但到最后,你真的需要能规模化。"

"Automating the process is the product."

"把流程自动化,这件事本身就是产品。"

"Okay, sales and customer success. The biggest challenge facing founders here is what I'll call the early demand trap."

"好,来说销售和客户成功(customer success)。创始人在这里面临的最大挑战,我把它叫做"早期需求陷阱"(early demand trap)。"

"It's easy to sign up a lot of pilot customers when you're just starting out and have nothing, but it can quickly overwhelm your ability to serve them, and you won't be able to build the product to scale."

"刚起步、手里什么都还没有的时候,签下一大堆试点客户(pilot customers)很容易;但这会迅速压垮你的服务能力,让你没法把产品做到可以规模化。"

"You'll be stuck using humans. It is a literal trap."

"你会被困在"只能靠人力硬扛"的状态里。这真真切切就是个陷阱。"

"Our advice here is to cap your first pilot customers to a small handful. Resist the temptation to sign too many too quickly."

"我们的建议是:把最早的试点客户数量控制在很小的一撮。要忍住"签太多、签太快"的诱惑。"

"Assuming you avoid this early demand trap, pre- and post-sales looks pretty different, too. You have to sell outcomes, not seats or tokens."

"假设你躲过了这个早期需求陷阱,售前和售后也会长得很不一样。你要卖的是结果(outcomes),而不是席位(seats)或 token。"

"The pilot is the product. For the first handful of customers, don't try and standardize too early. Use those pilots to learn."

"试点本身就是产品。对最早的那几个客户,别急着标准化。要用这些试点来学习。"

"Find the spots where AI gives you unique leverage versus spots where you're just automating something obvious."

"找出哪些地方 AI 能给你独特的杠杆,又有哪些地方你只是在把显而易见的东西自动化。"

"Build the product accordingly, and do it fast."

"照着这个来做产品,而且要快。"

"Pricing is harder than traditional software, cuz you're not competing with other software providers, you're competing directly with the cost of labor, internal or outsourced."

"定价比传统软件更难,因为你不是在跟其他软件供应商竞争,而是直接在跟人力的成本竞争——不管这人力是内部的还是外包的。"

"A few options here on pricing. There's per unit pricing, so per return, per claim, per loan. This is the cleanest, it's the easiest to explain."

"定价有几种选择。一种是按单量计价(per unit)——按每一份报税、每一笔理赔、每一笔贷款收费。这种最干净,也最容易解释。"

"There's also outcome-based pricing. This aligns incentives beautifully, but it can be harder for you to forecast in your business."

"还有一种是按结果计价(outcome-based)。这种方式在利益对齐上非常漂亮,但会让你更难预测自己业务的收入。"

"Panacea prices on the completed consultant study versus hourly, which is the norm in the industry."

"Panacea 是按"完成一份顾问研究"来计价的,而不是行业惯例的按小时收费。"

"There's definitely two pricing strategies to avoid. Cost-plus pricing captures your upside permanently. Don't do it."

"有两种定价策略是绝对要避开的。一是成本加成定价(cost-plus)——它会把你的上行空间永久性地锁死。别这么干。"

"Straight-line undercutting makes your work seem cheap and potentially low quality. Price on value."

"二是一味压价抢单,这会让你的工作显得廉价、甚至像是低质量。要按价值定价(price on value)。"

"Next, the P&L or the profit and loss statement. This is where these companies live or die, so let's do a quick walk-through."

"接下来,P&L,也就是利润表(profit and loss statement)。这类公司的生死就在这上面,所以我们快速过一遍。"

"If you haven't stared at one of these financial statements before, you're not alone among founders, and that's okay."

"如果你之前从没盯着看过这类财务报表,你不是唯一一个这样的创始人,没关系。"

"The general structure is revenue minus cost of goods sold gets you to gross profit."

"大致结构是:收入减去销货成本(cost of goods sold),得到毛利(gross profit)。"

"Gross profit minus operating expenses gets you to operating income. Let's dive into each of these in the context of AI services."

"毛利再减去运营费用(operating expenses),得到营业利润(operating income)。我们放在 AI 服务的语境里,逐项来看。"

"Revenue. Ah, the easy part, relatively speaking. You will be able to sign contracts. Can you deliver on them repeatedly? I don't know."

"收入。啊,相对来说这是简单的部分。你会签下合同。但你能不能一次又一次地把它们交付出来?我不知道。"

"That depends on your product and your process. Eventually, you want smooth and predictable growth."

"这取决于你的产品和你的流程。最终,你想要的是平滑、可预测的增长。"

"A great product process is going to smooth out that lumpiness, but in the early days, it'll be spiky on a monthly basis. That's okay."

"一套优秀的产品流程会把那种忽高忽低抹平,但在早期,按月看它会是尖刺状、上蹿下跳的。这没关系。"

"Cost of goods sold or COGS, obsess over this from day one. There's three main components here."

"销货成本,也就是 COGS——从第一天起就要死盯它。这里有三大组成部分。"

"There's model costs, hosting costs, and those humans in the loop. All three of them need a number, a trend line, and someone who owns them."

"分别是:模型成本、托管成本,以及那些闭环里的人。这三项每一项都需要有一个数字、一条趋势线,还有一个专门负责它的人。"

"Be deeply suspicious of zero margin or negative margin pilots. They're fun to learn from, but it's really dangerous to get hooked on those."

"对零利润、甚至负利润的试点要高度警惕。从中学习很有意思,但一旦上瘾就非常危险。"

"The core bet here is the more the product is built, the lower the COGS, the better the gross margin. I call this AI operating leverage."

"这里的核心赌注是:产品做得越到位,COGS 越低,毛利率越好。我把这个叫做"AI 运营杠杆"(AI operating leverage)。"

"Okay, OPEX. This includes research and development costs, so building the product, sales, and general administrative, which is like finance, legal, exec salaries."

"好,OPEX(运营费用)。这包括研发成本——也就是做产品——销售费用,以及一般管理费用,比如财务、法务、高管薪水。"

"Pretty standard stuff here. Operating income. This is the revenue minus the COGS minus the opex."

"这些都挺标准的。营业利润:就是收入减去 COGS 再减去 OPEX。"

"You will be judged on your operating income in these businesses faster than you might expect."

"在这类生意里,别人拿营业利润来评判你的速度,会比你以为的快得多。"

"Net income, this is the operating income less taxes and interest. That's less important in the medium term."

"净利润:就是营业利润再减去税和利息。这个在中期没那么重要。"

"So, here's the actual P&L opportunity for these businesses. Traditional services firms top out around 30% margins."

"那么,这类生意在利润表上真正的机会在哪?传统服务公司的利润率大概到 30% 就见顶了。"

"Pure software and agent companies have more margin but often smaller TAMs."

"纯软件公司和 agent 公司利润率更高,但市场总量(TAM)往往更小。"

"The bet on these services companies is that that AI operating leverage gets you closer to software margins."

"押注这些服务公司,赌的就是:AI 运营杠杆能把你的利润率拉到接近软件的水平。"

"Say 50% plus on a market that's two to three times bigger than software."

"比如做到 50% 以上,而且是在一个比软件大 2 到 3 倍的市场里。"

"You don't need to be there right away, that's okay. But, the trajectory has to be believable."

"你不需要一上来就到那个位置,这没关系。但这条通往那里的轨迹,必须让人信得过。"

"Okay, last but not least, don't try to buy your way in."

"好,最后但同样重要的一点:别想着靠"买"来进场。"

"There's a temptation we've seen, especially among founders with some operating background, to try and buy an existing services business, add some AI on top, short circuit the revenue."

"我们见过一种诱惑,尤其是在有一些运营背景的创始人身上:想去收购一家现成的服务公司,在上面加点 AI,抄近路把收入搞起来。"

"This is generally a trap. There's one decent reason to do it. You need a regulatory moat fast. Insurance licensing, for example."

"这通常是个陷阱。只有一个还算说得过去的理由:你需要快速拿到一条监管护城河,比如保险牌照。"

"But, otherwise, this almost never works. So, why is that? You just can't acquire product market fit."

"但除此之外,这几乎从来行不通。为什么?因为产品市场契合(product market fit)是买不来的。"

"Legacy service businesses are, you know, legacy. They have different expectations on metrics, hiring, and performance."

"传统服务公司嘛,你懂的,就是"遗产"级的老家伙。它们在指标、招聘、绩效上,有一套完全不同的预期。"

"Adding AI on top of that doesn't immediately change any of those realities. Building is almost always better than buying."

"在这上面加层 AI,并不会立刻改变这些现实中的任何一条。自建,几乎总是比收购更好。"

"Okay, to recap, AI services companies present an incredible opportunity for today's founders. But, these are fundamentally different startups to build."

"好,总结一下:AI 服务公司为今天的创始人带来了一个惊人的机会。但要造的,是一种根本不同的创业公司。"

"If you avoid the traps and focus on the process as the product and the product as the process, you have a chance to create a generational company."

"如果你能避开那些陷阱,并牢牢抓住"流程即产品、产品即流程"这一点,你就有机会造出一家足以定义一代人的公司。"

"We're excited to see what you build and hope you apply to YC."

"我们很期待看到你会造出什么,也希望你来申请 YC。"