Creator perspective · MoneyHormozi's 'shifting from local to global' technique: expand the buyer's consideration set by introducing more variables than they had thought about, since you know more about the domain than they do. The point of a good sales process is educating the buyer, and you should make clear they'd be getting way less for less, not the same thing for less.
Alex Hormozi · Give Me 42 Minutes, I'll Show You How To Sell Like The Top 1% | Scale or Fail Episode 1
Principle · SocialThe coach advises switching from topic to topic until you find a topic you both click on, then talking about your experience in that topic to start connecting on a deeper level. He illustrates this with the client connecting over reading and poetry since she's a literature teacher.
Coach Kyle · Will He Get Rejected After Approaching This Woman? (Infield)
Principle · MindHormozi argues that a single betraying punishment event can undo all the prior rewards and reinforcements in a relationship. He illustrates with marriages where someone trusted for 20 years then did one thing and could never be trusted again. Breaking your word doesn't kill the relationship immediately — like a tree cut off from the vine, it stays visibly standing but is dead on the inside. Truth functions as the stream of nutrients in a relationship, and it gets cut off the moment someone betrays somebody else.
Alex Hormozi · How The Top 1% Actually Think About Trust
Principle · MindTo build trust with someone, give them a little bit of information at a time and increase it as they prove reliable — if they don't do anything bad with it, share more. To deepen a relationship, increase shared context by volunteering information about yourself: every single thing you volunteer gives the other person context.
Alex Hormozi · How The Top 1% Actually Think About Trust
Principle · MindTo build trust with someone, share something they could use to hurt you — make yourself punishable. When you trust someone with your kid, you give them two things you value: your kid and your word, and the environment punishes failure to keep the word — not the person who asked. Sharing something that could be used to hurt you, while trusting the recipient not to, also gives everyone else the opportunity to earn more of your trust. The speaker notes using this himself, e.g., before giving a presentation to his company.
Alex Hormozi · How The Top 1% Actually Think About Trust
Principle · MindA single punishing event — a betrayal — will literally undo all the reward and reinforcement built up beforehand. When someone betrays your trust after years of reliability, the betrayal erases the accumulated trust. The relationship doesn't die immediately, but it is cut off from truth like a tree cut from its vine. This is why breaking your word for a short-term benefit doesn't just harm the relationship — it likely ends it.
Alex Hormozi · How The Top 1% Actually Think About Trust
Principle · MoneyHormozi insists the team should know exactly how much money they make on every event so that at year end they know precisely what they earned. Without per-event profitability tracking, you end the year not knowing how much money you made.
Alex Hormozi · How to Get Rich for Service-Based Businesses (Thank Me Later)
Report · MoneyThe whole point of Cloud Managed Agents is that you can do long-running, asynchronous tasks — not just for minutes, but hours, days, weeks, and even months.
AI Engineer · Build a Research Agent with Claude Managed Agents (Part 2 of 4) — Anthropic
Report · MoneyThe whole point of Claude Managed Agents is that you can do long-running, asynchronous tasks — not just for minutes, but hours, days, weeks, and even months.
AI Engineer · Build a Research Agent with Claude Managed Agents (Part 2 of 4) — Anthropic
Report · MoneyThe first workshop step makes sure prerequisites are set by copying .env.example into .env.local. Attendees paste their API key and put their name and email into .env.local as their Edgar identifier, because the SEC wants to know who is calling their API.
AI Engineer · Build a Research Agent with Claude Managed Agents (Part 2 of 4) — Anthropic
Principle · MoneyTo make a content business work, you have to be more than just content. Google traffic alone isn't going to sustain a business; you need other avenues to generate revenue and expand your reach.
Ahrefs · The Greatest Google SEO Recovery of All Time?
Creator perspective · MoneyFlywire's adoption advice: start small by picking a repetitive, well-defined task and proving it works before scaling; use the model only where it adds value while everything predictable stays as code, which keeps workflows reliable.
AI Engineer · Agents in Orbit: Workflows, Not Autopilot — Eduardo Simon & Javier Casal, Flywire
Creator perspective · MoneyKevin explains that we no longer have as many long-running applications; instead there are specialized agents that are sandboxed, and their demands on their compute environment differ greatly. Agent workflows demand different things: how much memory or CPU the sandbox has, and the permissions available to processes and file systems.
AI Engineer · https://www.youtube.com/watch?v=k72eLcYpVys
Principle · MoneyNnenna Ndukwe recommends maintaining a hierarchy of code-quality rules: general engineering standards that apply to any language at the top, then language-specific skills (e.g., all Python repos) below. The more general a skill is, the more repos need access to it — decide relevance based on your use case. You can scope rules or skills to very specific repositories yourself, or let the tool infer it. She also uses a combination of different agent skills depending on the repos, programming languages, and frameworks involved, and has coding agents build skills themselves if they're not publicly available.
Endorsed; the generality heuristic carries slightly lower confidence.
AI Engineer · How to Build Quality Gates into Agentic Coding Workflows — Nnenna Ndukwe, Qodo AI
Principle · MoneyNnenna Ndukwe recommends rerunning the verification command after fixing coding changes found in PR review, not just stopping at the local run. Because the code has changed, a fix can raise something else, so any automation should include verification sections again. Whether verification runs automatically after fixes depends on your agent, but you can prompt your agent to run verification regardless.
Endorsed as best practice 'just in case'; no strong claims.
AI Engineer · How to Build Quality Gates into Agentic Coding Workflows — Nnenna Ndukwe, Qodo AI
Principle · MoneyNnenna Ndukwe believes in running local code review workflows regardless of which tool you use (Codex, Claude Code, Qodo, etc.), always before another developer has to see your code. The point is to catch small changes earlier rather than later and mitigate issues popping up in formal review.
Strongly endorsed; no stated uncertainty.
AI Engineer · How to Build Quality Gates into Agentic Coding Workflows — Nnenna Ndukwe, Qodo AI
Principle · MoneyNnenna Ndukwe recommends that when working on a team, skills and rules should ideally live outside your local repo unless you have a good use case for containing them there. She suggests using a context management layer (like Unblocked) as a separate place to manage that context, which you can pull down wherever you are, and practicing having rules, skills, and other engineering standards stored elsewhere so they can be pulled into relevant workflow stages.
Endorsed with a stated exception (good use case for local containment); the Unblocked mention carries lower confidence.
AI Engineer · How to Build Quality Gates into Agentic Coding Workflows — Nnenna Ndukwe, Qodo AI
Creator perspective · MoneyThe source splits inference into two major phases: pre-fill and decode. During pre-fill (the first couple of seconds where ChatGPT returns nothing), the system computes K and V vectors for every prompt token and produces the first output token; this phase is very compute-bound because it can be parallelized extremely. After the first token, decoding happens one token at a time, reusing the KV cache built in pre-fill, and is more reliant on memory because each decode step rereads the entire KV cache.
None significant.
Also discussed by Y Combinator (unclassified source — not counted until classified). Not counted as independent sources.
AI Engineer · https://www.youtube.com/watch?v=_cRgl3maiT8
Creator perspective · MoneyThe source explains that LLMs generate tokens one at a time in an autoregressive loop, and to produce each additional token all model weights must be read from memory into compute. As a result, up to 90% of the time during token generation the compute sits idle waiting on memory. This is the foundational explanation for why inference is memory-bandwidth-bound.
The 90% figure is presented as an upper bound ('up to 90%').
AI Engineer · https://www.youtube.com/watch?v=_cRgl3maiT8
Principle · MoneyThreat modeling must be repeated as the architecture evolves, because apps and agentic systems change over time; it is not something done once at creation.
AI Engineer · AI Security Engineer Foundations + Certificate — Javier Garza, Snyk