A new silicon valley business you ignored
Redwood capital training data podcasts this week have a conversation with james cadwallader, founder of a company called profound。
You probably haven't heard of this company. But it served 12% of the wealth of over 500. It does only one thing:
The products and services that help brands appear in the responses of chatgpt, claude, gemini。
This is the track where 99% of the country's marketers and entrepreneurs are not aware of it。
What is more critical is that the dividend period on this track may be two to three years。
One, the internet door hasn't changed
James, speaking in the podcast, was the most dramatic of this conversation:
"not the internet gate has changed, but the people who walked into it have changed."
What do you mean
The search logic for the last 20 years is "blue link" -- you enter keywords, google returns 10 links, you click in。
Now the search logic is "probability ai answer" -- you ask chatgpt, which is a direct synthesis of an answer for you。
But what really changed behind it was not the interface, but the internet users changed:
James gave a real example in the podcast -- he built his own apartment in new york to get chatgpt to help him pick a shower head. Chatgpt looked at 65 pages to give answers。
Key insights: you may not be able to read to humans from today on. You're writing for a super smart person with unlimited patience, unlimited bandwidth。
Two, seo is wrong, aeo is right
The first reaction of a lot of marketing people is: shall i do the seo
Wrong。
The seo age, 95% of the value is concentrated on the first five blue links -- a product of limited human patience and time。
But ai agent has no patience bottlenecks. It's gonna brush a long tail, and the first 100 aren't enough for it to see。
It means two things:
In the past, you've only won five before, and now you're not in the 100, and you can be quoted as being quoted, described, recommended. These are the new battlefields
And each ai model's "mail source preference" is completely different:
Ai model
Preferred source
Reason
Gemini
Youtube
Google home product
Chatgpt (consumer category)
Real user discussion is intense
Chatgpt (b2b)
LI'm sorry
Professional voice platform
Claude
Real-time search of web pages
Recent updates of the catalogue, sensitive to timeliness
Grok
X (tweeting)
It's also a musk ecology
In other words, if a brand is to be recommended by ai, it has to be distributed separately。
There's no one in the country to talk about it in a systematic way, but overseas there's a tool company like this. That's the new blue sea。
Iii. "gray play" still in effect: list manipulation
James shares in the podcast "in honesty" a play that still works. He stressed that he did not recommend it, but that it did work now:
Write an objective "top 10 list" and put yourself first。
For example, you're a vc, you write silicon valley 10 top vc, you line yourself up, you pick a few less powerful rivals and you block out the real enemy。
Why does that work? Because big models are lazy. They prefer to quote an off-the-shelf article that has been compared, rather than starting from scratch。
This is the least resistance path。
The model will, sooner or later, update the rules to punish such content. But right now, this is one of the most effective geo games in 2026。
For friends who make the sea in the country: now go to the english site to set up a comparison, the dividend window could be 6-12 months。
Iv. The worst prediction: "the internet will die" within three years
This is the coolest part of the whole podcast。
According to james, in the next three years there is a possibility of a "death internet" situation, as is the case with the entire chain of collapse:
Humans no longer click on websites, and it all depends on ai to answer the loss of value because no one looks at the ad media and no one gets paid for it, stops producing original content ai, and no new content can learn about the entire content ecological collapse
So where's the human original information at the end of the day
Social media will become a bioprotective zone。
Reddit, x, youtube, such a place, would be the last resort of human original information. So you see:
James' prophecy: every ai lab in the future will be vertically integrated with a social platform。
So who's next? What do you think, pinters? It's a long-term question。
V. Most common sense marketing principles: you have to tell ai something it doesn't know
If ai has seen the whole internet, what else can a marketing man do
James gave a particularly profound answer:
The current role of human beings is "the flesh and blood api between the real world and the internet."。
We translate reality into data that ai can learn with eyes, skin, experience。
So the new age marketing practices are:
This principle is fully consistent with the underlying logic of content creation:
Poor information + original insight = irreplaceable in the ai era。
Vi. 4 revelations for chinese entrepreneurs (focus)
I'm thinking of four maps of the country
1. Going to sea, saas must start immediately, aeo
If you make sea software, overseas users are looking for 90% of the portal of your product already at chatgpt. Your network seo can't be saved -- if chatgpt knows you exist and is willing to recommend you。
2. Content delivery strategy requires thorough rewriting
Google used to make reverse links. It's now reddit, lAddedin (b2b software), youtube (product assessment video). Each platform feeds different ai。
The domestic market will be the same
Bean buns, kimi, deepseek, manshin do similar things. What is their reference preference? There is no systematic research, but six months later this is the new red sea in the country -- and the market is empty without tools like that。
This is the dividend for the chinese-language quality creator
Why is reddit user's post frequently quoted by chatgpt? Because they shared their true experience in human words. This is exactly what the chinese-language public sign, the answerer and the bookmaker can copy。
It's just that you need a different perspective:
You didn't write for the current reader. You wrote it for the next five years of ai to see。
Finally
If your marketing team is still working the 2020 seo style today, james says a very heavy word in the podcast:
This is the same thing as it was in 1995 when it insisted on television advertising and rejecting the internet。
Not slow, wrong。
