Yan Shi retweeted
A toothpaste company has quietly killed the entire market research industry and nobody is talking about it.
Colgate published a paper showing you can predict real purchase intent by simply asking LLMs to roleplay customers.
And this is beyond insane.
If you ask an AI, "Rate this product from 1 to 5," it gives safe, middle-of-the-road garbage.
So researchers invented a method called Semantic Similarity Rating (SSR).
Instead of asking the AI for a number, they asked it to roleplay.
They gave the LLM a demographic profile. They showed it a product concept. And they asked it to write down its raw, unfiltered thoughts.
Then, they used a semantic model to translate those written thoughts into a numerical score.
The results are staggering.
Tested against 57 real personal care product surveys and 9,300 human responses, the synthetic consumers hit 90% of human test-retest reliability.
They perfectly mirrored how different age brackets and income levels react to price changes.
And they provided detailed, qualitative feedback that was deeper and more critical than what actual humans wrote.
This destroys the economics of traditional market research.
You don't need to wait a month to see if a product will sell.
You can simulate 1,000 hyper-targeted customer interviews overnight.
You can A/B test pricing across every demographic instantly.
Yan Shi retweeted
being on the phone is probably the most important skill a salesperson can have right now
more of the companies I speak to are going back to building proper cold calling teams, and letting AI and automations handle the backend stuff for them
Yan Shi retweeted
HubSpot just brought their co-founder Brian Halligan back as CEO and the first thing on his list should be Salesforce going headless
you can now run Salesforce through CLI commands and MCP without ever opening the UI. that's going to absolutely cook HubSpot unless they get there too
the biggest issue people have with HubSpot and CRMs in general is the UI. it's too complicated, so you end up paying an expert to come in and run it for you. no one actually wants to do that
but the upside of a really good CRM is huge, and if an agent can run it, the expert bit goes away
Yan Shi retweeted
Yes, kudos to HubSpot for backpedaling on the “Customer Harness” thing.
It’s maybe the most high-profile example of a point I keep making: people in tech are too close to it.
Everywhere I look, founders wind-whipped by AI are losing touch with “normal” businesses.
Yan Shi retweeted
Replying to @rjs
I’m reading Daniel Jackson’s “The essence of software” and finding his “concepts composed via synchronization” model very promising
Yan Shi retweeted
A huge percentage of my life is spent avoiding crowds. Avoiding fashionable ideas, investing in underpriced assets, finding cool places that aren't deluged with tourists — they're all variants of the same thing. Whenever I find myself in a crowd, I feel something is wrong.
Yan Shi retweeted
npm downloads is a good metric for attention, and might not necessarily correlate with monetization potential, very likely quite the opposite, people will connect it with free and you'll either race to acquisition to someone like cloudflare, or race to death
Yan Shi retweeted
Project management should be 80% design, 10% logs, 10% tasks.
The clearer the design is, the fewer tasks you need.
Some logs are necessary for capturing why the design is the way it is.
Yan Shi retweeted
发现一个比哥飞那种找词还要快的Web出海的玩法:
flippa.com/ 上都是卖网站的,上面都露了网站多少钱,每个月赚多少钱,你让你的AI去复刻,至少赚钱是确定性的,不知道多少而已。
后面你把SEO做起来不就完事了吗,这不比找词确定性高多了吗?
Yan Shi retweeted
We’re really struggling to manage product development right now.
For years, we used our own take on Shape Up. But with the absolutely insane speed of agentic development, writing a proper pitch and setting up a project often takes longer than just building the thing.
With 10 things being built at once, keeping an overview and making sure the team still feels in control is getting harder and harder.
Yan Shi retweeted
The more I work with AI, the more I think @soleio is right:
“The future belongs to designers who build.”
It’s wild how much creative range opens up when you can actually make the thing yourself instead of translating it into a handoff for someone else.
No translation layer, no waiting for someone else to make it real. Just idea → prototype → iteration.
Yan Shi retweeted
Why HubSpot needed to get back to founder mode, in one slide:
Customers were mostly staying (101% NRR). But new ones fell by half in just 2 quarters.
Yan Shi retweeted
When coding agents start making design decisions, it's common for UI drift to appear. That’s when our team comes in to clean it up.
We take a thorough inventory, build a design system that both engineers and agents can use, and add guardrails to keep drift from coming back.
@buildkite and @currents_dev have already worked with us on this. Check out this service and book a call: evilmartians.com/services/ai…
Yan Shi retweeted
以前在美团实习的时候,翻到过美团创始人之一王慧文的一个观点: “中国的toB软件发展缓慢很大一个原因是中国企业分布是头部很大 尾部很多,但腰部企业很少。”
美团曾经收购过一家叫“天子星”的公司,是做餐饮连锁信息管理系统的,当时收购的一大原因是这家公司在头部商家的占有率最高(说人话就是头部餐饮公司选择天子星产品和服务的非常多)。
本着最头部的商家选了谁 谁家产品就更具有竞争力的原则,美团收购了它。
但是一收进来竟然发现天子星的财务状况非常糟糕,美团内部又调研了一圈,结果发现行业里财务状况好一些的都是头部占有率低的。
后来美团高层、战投部门、商分部门就开始复盘这件事,得出的结论是在餐饮行业里 头部商家的议价能力非常强,也会提很多定制化需求。
但是定制化需求是个无底洞,而且餐饮软件公司还不断有新公司进场,所以头部商家永远不缺公司给他们低价做定制化项目。
所以很多餐饮软件公司的玩法都是做完一个头部的标杆项目之后就去做腰部企业了。
但这种公司过的也很艰难,因为中国腰部企业太少了,饼就这么大,每多撕下来一点都很难。 看似不相关的问题,但实际上美团内部得出的这条结论对于相当一部分toB业务是具有普适性的。
飞书这样做toB生意的办公管理软件, 如果做大企业大公司,首先很多大公司内部很可能会有一个团队就在做类似的系统,说服人家抛弃旧的换一套新的系统很难。
其次,大公司难伺候,做大公司生意会陷入无穷无尽的定制化需求当中,但是“定制化”这件事是完完全全反规模效应的。
字节无法利用自身的流量优势和规模优势,高薪聘请的各种工程师和产品经理只会成为成本负担,做这种生意只会沦为高级外包。
如果做小企业小公司,首先小公司就压根没有用你这套系统的需求,其次小公司也不具备支付能力。 那做腰部企业呢?对不起,中国的腰部企业太少了。
不够供养字节飞书如此之大的体量。



















