Unfortunately, SEO Just Changed Forever

Eric says marketing is moving beyond social and Google search as companies optimize to appear in large-language-model results. Google search volume is rising, with users asking longer questions, but search-ad revenue reportedly increased only 12%, making investors worry that ChatGPT will displace Google. Eric argues Google is increasingly being used like ChatGPT and has not yet fully monetized its AI answers; more ads may eventually appear. The larger shift, he says, will come when agents such as Dots, GrokBot, and Muse search on behalf of people. A person may make two or three queries, while an agent could run 40 or 50 granular searches before selecting a company.

Eric illustrates the shift with his co-founder Mike, who used Grok as a digital personal assistant to find water-heater companies willing to visit his home and provide quotes. Mike specified his price range, and the agent helped him choose among the results. Eric says companies currently optimize for search and social discovery but are not preparing for a world in which AI conducts most searches behind the scenes. Total search activity may rise even as human searching falls. Businesses that are not structured so agents such as Muse, Dots, or GrokBot can understand and recommend them may not be selected.

Eric describes using his ChatGPT-based assistant, which he calls Neil, while organizing a global AI summit. The assistant noticed a problem involving speakers who were not registered for city-of-light access and offered to handle it. It then contacted Carol as “Doodoo, Eric’s AI assistant,” asking who Eric should contact to escalate the request. Eric says he is increasingly trusting agents with tennis bookings, dinner reservations, and other complicated tasks, and expects to delegate more as confidence grows. Neil jokes that Eric should rename the assistants: an AI representing him as “Doodoo” could undermine his image.

Eric says GrokBot is still valuable, but his usage appears to have been cut roughly in half, while API overage bills have suddenly increased. He and Neil attribute this to compute economics: Elon Musk can sell scarce computing capacity to hyperscalers for more money, so reducing consumer access may be commercially rational. Eric also describes a genealogy bot that used GrokBot to uncover previously unknown details about his grandfather, including government, wartime, newspaper, and Taiwanese stock-market activities. Their practical advice is to use agents, but name them carefully and recognize that providers may restrict usage as costs rise.

Eric explains a chart about Anthropic’s internal AI model research and development after Neil initially finds it confusing. The chart shows AI moving from a minor role, to assisting researchers, to collaborating with them, and then leading work. By August 2026, Eric says AI was leading about 26% of Anthropic’s work, while the “minimal AI” category was falling. They connect this to recursive self-improvement: AI systems doing work that improves future AI work. Eric argues that people who say AI’s return on investment has not appeared may simply be behind frontier organizations such as Anthropic, where the change is already visible.

Eric presents Andreessen Horowitz data sourced from JPMorgan and Adobe comparing AI-referred shoppers with other online shoppers. AI-referred traffic reportedly has 60% higher conversion, 53% higher revenue per visit, and a 34% lower bounce rate. The year-over-year growth of AI traffic slowed from 269% in March 2026 to 130% and then 62% by July, but it remained positive. Neil notes that conversion uplift rose from 42% to 60%, while revenue-per-visit uplift increased from 37% to 53%. He cautions that agencies he knows are not seeing such dramatic increases, though they do see AI traffic convert unusually well.

Eric compares the YPO Global Business Summit in Istanbul to Coachella or EDC for people who love business. Rather than attending talks, he spent about eight hours in “needs and leads” sessions, where groups of roughly ten people each stated what they needed and others indicated whether they could help. He says he generated around 40 leads, including introductions involving acquisitions, animal-lifespan technology, medical devices, cancer research, and investors. The summit cost about $5,000, though Eric combined it with his own AI summit to reduce the overall expense. He says the event’s direct, transactional format made it especially valuable.

Eric tells Neil that a mutual friend initially planned to invest $250,000 in a company but increased the amount to $1 million after a further five-minute conversation. The founder had previously built a company generating hundreds of millions in revenue and explained that a personal loss motivated his current work. Eric says the decision rested on both the compelling story and the founder’s demonstrated ability to execute. The founder was not trying to raise a full million; he mainly wanted suitable people on the cap table to support a future public listing. The summit’s 700 YPO members gave him access to that network.

Neil breaks down why high-interest acquisition financing is dangerous. If a buyer purchases a company generating $1 million in profit for six times earnings, the price is $6 million. At 15% interest, $900,000 of annual profit goes only toward interest, leaving little to reduce principal, assuming profit converts fully to cash. At roughly 6% interest, the annual cost is about $360,000 to $400,000, leaving $500,000 to $600,000 before taxes for repayment. Neil says his own facility costs about 2% plus SOFR, roughly 6% at the time, and allows interest-only borrowing, making two or possibly three times leverage more attractive than five times.

Eric says the summit rewarded people who stated a precise need and gave introductions without expecting an immediate return. He connected a television company offering performance-based advertising with several prospects, because its offer was simple: it would fund the TV campaign first and get paid based on performance. He also helped match people seeking influencer marketing, AI transformation, and agency-sale advice. Neil agrees that giving can produce relationships naturally, although he does not help every competitor. Eric contrasts the strong TV pitch with “we have a staffing company in the Philippines,” which became more compelling only after being reframed as access to top talent from companies such as American Express and Visa.

Eric and Neil discuss the U.S. midterm results as a marketing lesson rather than political advice. They argue Democrats won because their message focused on high gas prices, high prices generally, and the ability to pay bills now. Republicans emphasized AI, data centers, future competition with China, and the national deficit, but those themes felt remote to people worried about food, rent, jobs, and bills. Neil applies the same principle to agency sales: clients do not primarily want higher rankings, more ChatGPT visibility, or more leads; they want increased revenue, profit, promotions, and security. Marketing should connect tactics to the buyer’s economic outcome.

Eric compares two marketing agencies at a needs-and-leads table. One offered a free marketing audit, attracting only one or two interested people out of ten. Eric described his agency as building AI workflows for scaling generative-engine optimization, SEO, advertising creative, cold-email infrastructure, and AI-native recruiting, then asked for pilot users; eight or nine people raised their hands. He tested an even narrower pitch focused only on an AI-native talent workflow, and nearly everyone expressed interest. Eric says the underlying service may be similar, but a specific problem, offer, and outcome are easier to remember than a broad audit.

Neil shares advice from Yaniv, the founder of Nextiva, who raised $200 million from Goldman Sachs Asset Management in 2021 at a $2.7 billion valuation after building the company without external funding. Yaniv told Neil that if life is already financially comfortable, taking five or six times leverage may create unnecessary stress. Eric recalls his own 2021 agency acquisitions, which used revenue-based financing but failed to execute as planned; the debt still came due, creating substantial pressure. He says the experience taught him that calculated risk is different from aggressive leverage. Neil adds that a bank told him only one in 400 loans could fail before its low-rate lending model lost money.

Eric describes agency owners who accepted private-equity funding to build acquisition platforms. One founder now wants out because the timetable is approaching the point when the agency must be sold again, often to another private-equity firm. Although the investors may offer more money to retain him, he no longer feels able to operate naturally or pursue what he wants. Neil recalls that Eric previously advised him not to accept celebrity investors, even for small stakes, because outside owners could create unwanted reporting and decision-making obligations. They conclude that founder freedom and the ability to choose one’s direction can be worth more than a larger valuation.

Neil and Eric say they dislike scripted short-form content and prefer conversational podcasts or screen-shared long-form videos. They argue that creators should choose subjects they genuinely care about because audiences can detect boredom; Neil is far more engaged discussing new agent optimization than basic SEO. Eric cites Steven Bartlett, who declined celebrity interviews that would require hours of preparation simply because they promised large audiences. Neil’s lesson is to interview for impact and useful knowledge, not views alone. Eric closes with a sales rule from the summit: make one clear offer or request in one sentence. Multiple unrelated asks make people forget what they are supposed to help with.