Why You Should Give Away All Your Workflows

Eric says HighLevel CEO Sean told him the company was “giving AI for free,” arguing that businesses should do what is best for customers even when they must absorb the expense. Eric sees this as unusually bold because a company with many users cannot easily predict the cost of unlimited token usage. Neil agrees that funded companies usually make expensive-looking bets only when the expense is bounded; Eric says HighLevel’s decision is riskier because its cost ceiling is unknown.

Eric describes a debate with a new managing director and his CTO over whether their agentic workflows should be protected as intellectual property. He argued that anything built now may be outdated within three to six months, so the company should give away workflows, AI-visibility reports, templates, and bots to build mindshare. Most users will use them themselves, but a smaller group will pay for implementation. Their bots also connect to the SingleBrain gateway, creating a path to paid services and workflow support.

Eric says an AI-visibility report that might once have cost $1, $2, or even $10 now costs about 18 cents, so 10,000 free reports would cost roughly $1,800. Neil warns that unlimited AI usage is dangerous when a company does not know its cost cap. Eric counters that future demand will require more compute and electricity, while hardware prices are already rising: his DGX Spark units went from $4,000 to $8,000. He believes companies are giving AI away now to secure usage before competitors copy them.

Eric compares HighLevel’s free-AI decision with much larger infrastructure bets. Meta executive Alexander Wang argued that the company had to invest in data centers even without knowing exactly what the spending would produce; CFO Susan Li repeatedly warned that failure could be very bad for the company, while Mark Zuckerberg accepted the uncertainty. Eric says Anthropic also failed to build data centers aggressively enough and benefited from a SpaceX deal. He considers data-center spending even scarier than HighLevel’s bet.

Using Andrej Karpathy’s example, Eric says only about 0.2% of the world—roughly 20 million people—currently use AI in the intensive way he and Neil do. Of 3.55 billion employed people, he estimates about 6% own companies and only 4% of those employ at least 20 people, leaving roughly 17,000 AI-native firms today. He also cites executive discussions about whether transformation is worthwhile and warns that top talent may leave companies that cannot offer an AI-native environment.

Neil jokes that Scale’s founder could tell the CFO not to worry because he had $14.3 billion in company cash, though Eric clarifies that this was not actually said. The reported $14.3–$14.8 billion payment was split among the founder, co-founder, and investors and was paid in cash rather than stock. They agree the founder was already financially comfortable, but Neil distinguishes paper wealth from cash and says being a billionaire is still preferable. Eric links this appetite for risk to founders such as Mark Zuckerberg and Elon Musk, who keep making bets they no longer need to make.

Eric says a fully loaded new touchscreen MacBook Pro was quoted at $14,000, while a fully loaded Mac Studio with 512 GB of RAM would cost about $21,000; Neil notes the laptop starts at $2,899 and jokes that this is an expensive microphone. They say Apple is marketing to technophiles who want the newest hardware, even though the touchscreen adds little. They then discuss a Chinese glass company whose unbreakable window survived Paris riot attacks; a social post showing chairs and kicks against it generated about 22,000 views and free advertising.

Neil says his bullet-resistant home windows cost roughly $200,000–$300,000 extra, because he prioritized protecting his family after protests and riots near Beverly Hills; he declines to explain what other protection he keeps. Eric then reports that a security executive friend is overwhelmed by inquiries after “Black October” hacks. The friend recommends password managers with unique passwords, preferably passkeys, plus strong two-factor authentication. Eric’s marketing advice after a breach is to disclose what happened, who was affected, and what it means; hiding the incident creates more anger when it eventually emerges.