Sachin:We are in the midst of this mega trend and how we as consumers, as businesses, search, discover, and interact with information online... It's not just ChatGPT, it's ChatGPT, it's Gemini, it's Grok and a whole long list of others.
James:Going chronologically... this is early 2024... this is when ChatGPT wasn't web connected... Perplexity were the ones who had the idea of okay we won't use the LLM to answer the user's query directly. We will use the LLM to synthesize from the web and create an answer.
It felt like a streaming to CDs type moment... two things became clear: the first was this is going to happen. This is inevitable because it's better. You don't have to click. You don't have to browse a list of blue links. You can just ask AI a question and get the answer.
And then the second is that you would need to understand that the economics of search or information retrieval as a whole would change with this new paradigm because people don't need to click the links... you see this disintermediation between websites and the consumer.
Sachin:One thing you said to me when we met back in November... Humans won't create content for machines. Machines will create content for machines. And when humans need to orchestrate that, they're going to do it in Profound.
James:There's like two core beliefs that we need to be true. The first is that we have to be correct that every marketer on the planet is going to care about how AI talks about their brand, products, services, categories, competitors. I believe that. I think in 10 years that will only be more true.
That will be one of, if not the most important priorities for every marketer on the planet... And then the second is that you have to believe every company on the planet is going to use AI to create marketing.
这太有意思了。 这句话可能是对 AI 营销最精辟的概括。当内容创作成本趋近于零时,人类的工作从"创作内容"变成"编排机器创作内容"。Profound 想成为这个编排层——就像 Salesforce 成为销售的编排层一样。
---
## #03 质量是上下文的函数
Sachin:We're only a month into 2026 and I feel like the VC buzzword of the year is context. And I think you were actually probably the first person to say it to me which is content cannot exist without context and all that context already exists within Profound.
James:I think quality is a function of context. So I think the more context that you give the models the higher likelihood there is that it's going to create something good. If you don't tell the models what you want... you need to hook up lots of internal data sources, external data sources like very complex orchestrations of data input allow the model to do a good job.
天哪,这个框架太重要了。 "Quality is a function of context"——质量是上下文的函数。这不是简单的"提示工程",而是构建一个完整的数据基础设施:15亿用户提示、CDN 集成的网站数据、竞品数据,所有这些上下文输入模型,才能产生真正高质量的营销内容。
---
## #04 18岁上门收黄金的经历
Sachin:I'm going to ask you to share the story, but I also know from getting to know you, you're a fan of props. I got you a little gift...
James:(拿起金链子)My background is pretty unorthodox in that I didn't go to school... When I was 18, a friend and I were looking for a summer job. It was 2009. It was just after the GFC and the price of gold... had gone sky higher. And we built a business trading gold, buying and selling gold.
We knocked on people's doors every day and asked to buy their broken old scrap jewelry on their doorstep. There's nothing harder than knocking on people's doors... you get told to f off a lot and it builds very thick skin.
这段经历太有画面感了。 2009年金融危机后,18岁的 James 上门收黄金,被无数次拒绝。这种"脸皮变厚"的经历,这种对人性直接的洞察,可能比任何 MBA 都更有价值。他学会了"能量交换"——如果你紧张,对方会感觉到;如果你自信微笑,人们会回应。
---
## #05 "大多数成功创业者本质上是不安全且焦虑的"
Sachin:Where are you pulling the most inspiration from as you think about running Profound?