注册 X 十几年,一直潜水不说话,今天开始认真发。
简单说下我是谁:在大厂做了 10 年+产品经理,手里有国家发明专利。23 年出来创业做 AI 电商出图,做到淘宝服务市场类目第一,然后被免费的大厂卷得不想融了😂
团队散了,人还在 AI 里泡着,这两年靠帮企业落地 AI 吃饭。来这儿就想跟大佬们多聊聊,谁真用 AI 赚到钱了。
Delta, Chase and Amazon are blocking AI agents.
The war on agents has begun, and here’s how it probably plays out:
Stage 1: Companies block agents because agents skip the ads and upsells.
Stage 2: Agents learn to act like humans, and it becomes an arms race.
Stage 3: The market splits. Companies that make money on sales let agents in, companies that make money on data charge for access, and the giants build their own agents.
Stage 4: A trusted middle layer for agent identity and payments becomes the standard way in.
Stage 5: Companies that welcome agents start winning customers from the ones that block them.
Stage 6: The agent blockers come back and let agents in on their own terms.
Stage 7: Everyone starts competing for agents with better prices, faster checkout and SEO built for agents.
Stage 8: The ads come back, and companies pay to be your agent’s first recommendation.
Stage 9: The agent war ends, and selling to agents becomes as normal as having a website.
Postiz hitting $275k MRR so FAST is a great example of what I'm seeing: tools that agents use for things like posting, email, payments, phone numbers and browsing are going vertical.
A few other businesses I keep seeing go vertical:
- Data brokering. Selling rare/useful AND expert knowledge to labs and to companies training their own agents.
- Hosting for agents. Renting always on computers so people can run their agents 24/7. A lot of people want an agent working for them and have no idea how to set one up.
- Compute reselling. Buying GPU capacity in bulk and renting it out to smaller teams who need it for a few hours.
- Vertical harnesses. Agents built for one industry and one workflow, like insurance claims or construction bids.
- Skills. People are selling packaged skills that make an agent good at a specific job.
- Done for you agent setup. Local businesses are paying someone to set up agents for them, the same way they used to pay someone to build their website.
When you zoom out, almost all of these do one of three things. They sell to agents, they feed agents, or they house agents!
Claude 团队刚交底,他们内部九成代码已经是 AI 写的,合并量翻了一倍多,撤回重改的比例反而降了四成。
秘诀也挺实在,不指望单次写对,干脆派另一拨 AI 排队挑刺,一层层拿筛子过滤。
现在企业落地也是一模一样,写东西早就不是瓶颈,怎么低成本验收才是真门槛。
毕竟代码可以让 AI 替着审,真把线上搞崩了,可没法让模型去向老板检讨😂
Claude Code 团队公开了他们审代码的做法:代码大约九成由 Claude 写,六月到九月合并的修改翻了一倍多,需要撤回的比例反而降了四成多。
这期用一摞带洞的奶酪片讲这套做法:每道检查都会漏,叠起来就很难漏,底下还有一张网兜底。
片中回答五个问题:哪些修改还要人看,AI 审 AI 会不会没完,审查的 AI 漏了怎么办,人不读代码怎么懂代码,出了事算谁的。人不再逐行读代码,时间花在机器做不好的判断上。
视频1: 英文版
视频2: 中文版
We built Box Mount to allow developers to mount Box as a file system directly in any agent sandbox. As agents do more complex work, they need to be able to work with files and data just like a person would. The future is headless.