digest
Short Summary
Main recommendation: Replace yourself with AI before someone else does, by rethinking your workflow with "first principles," then build businesses that sell "impossible outcomes" and "sell to agents, not just people." The speaker argues that people who play offense by disrupting their own work and offerings will capture the next wave of wealth, while those who protect their current roles and businesses will be left behind.
Key Points
- Replacement happens at three levels: AI-first individuals who outproduce teams, AI replacing whole departments such as customer support, bookkeeping, and content, and AI making entire businesses unnecessary, such as translation agencies, basic tax prep, and design work.
- Historical examples are offered as proof that businesses which refused to disrupt themselves were overtaken: radio ignoring television, newspapers ignoring the internet, billboard companies ignoring Facebook ads, and Blockbuster ignoring streaming.
- Sell results that could not exist before AI, rather than cheaper versions of existing AI tools, since AI chatbots and phone assistants are expected to be commoditized within 18 months.
- Businesses should prepare for AI agents that research, compare, and buy on a user's behalf, which the speaker projects will become a 3 to 5 trillion dollar market by 2030 according to McKinsey.
- Clear, structured websites with plainly stated offers are easier for agents to read and buy from than vague "contact us for a quote" pages.
Details Worth Noting
- Concrete instructions:
- Write down everything you do in a typical week, focusing on workflow rather than job titles.
- For each item, ask: if starting from zero today, would I do it this way, or would AI handle it?
- Replace first anything AI can handle, such as email triage, negotiating, coordinating, writing in your voice, and researching.
- Keep only tasks that AI cannot do, described as "hard for computers, easy for humans," such as talking to people, negotiating feelings, humor, and nuance.
- To find a sellable problem, identify a painful problem, study how current providers solve it through interviews, secret shopping, or calling them, ask how AI could do it in half the time, twice as fast, or cheaper, and pre-sell it by describing it as if it already exists before building it.
- Structure your website so an agent can answer four questions: what you do, what it costs, what clients get, and how fast, with clear "how to get started" steps.
- Avoid heavy animations and vague marketing language, since an agent only needs clarity.
- The speaker says a two-person company reached a million dollars in run rate within a few weeks by building a solution that was previously infeasible, though he does not name it.
- He cites layoff reports, including a rumored 20% cut at Meta and a 40% AI-related cut at Block (formerly Square), both presented as unverified rumors/reports he heard.
- He cites Coinbase's "agentic wallet" feature as an example of a company preparing for agents that control spending.
- He reports having paid for 13 tools in 10 days, with an AI agent choosing and signing up for them.
Bottom Line
The core message is to act early and proactively: audit your own workflow, offer results that did not previously exist, and make your business legible to AI agents before competitors do.
