Principle · MoneyA successful brand has three pieces: getting the customer to the website, a great offer, and a great product. You need a great offer (website, pricing, maximizing the customer's visit) to get customers to actually buy, and a great product to get customers to come back. If you don't have all three, your business is cooked. An amazing creative engine doesn't matter if you have terrible LTV because your product sucks, and you won't get great creators to work with you if they don't genuinely like the product. Build a great product first, then a great offer, then a great sales team — doing all three creates a monster business.
Foundr · The NEW Creator Playbook that Took Me from $80K Debt to $250K/Day | Trybe
Principle · MoneyThe greatest thing he gained from Hudson was the creator playbook — taking people who would be great customers and turning them into amazing salespeople — like door-to-door sales except potentially selling to thousands or hundreds of thousands over content.
Foundr · The NEW Creator Playbook that Took Me from $80K Debt to $250K/Day | Trybe
Principle · MoneyHis entire thought process shifted to how to make everybody else as much money as possible, on the principle that helping other people make money makes it all come back to you.
Foundr · The NEW Creator Playbook that Took Me from $80K Debt to $250K/Day | Trybe
Principle · BodyTo develop a big vertical you have to be strong enough to put serious force into the ground. The coach explains that weightlifting movements and squats are used because they teach rapid hip extension, which transfers well to jumping. He also notes that doing hundreds of jumps alone does not develop a bigger vertical.
Garage Strength · I Trained 1000 Elite Athletes. This Is How You Jump Higher.
Principle · BodyThe coach lays out a four-part framework for developing explosive power in the vertical jump: getting stronger, getting more explosive with load, strengthening the tendons (which happens during squatting), and big-time plyometric development. He stresses that you have to get strong, do the jumps, do the little exercises to develop tendon strength, and then coordinate everything into the right program.
Garage Strength · I Trained 1000 Elite Athletes. This Is How You Jump Higher.
Principle · BodyDespite people calling the box jump stupid or cheating — the common criticism being that you're just jumping up and raising your knees without developing anything — the host considers it one of the best ways to develop your vertical jump. It's a way to learn and accelerate, perfect for someone just getting into jump training, and teaches how to put force into the ground rapidly; done the right way it's a great exercise.
Garage Strength · I Trained 1000 Elite Athletes. This Is How You Jump Higher.
Creator perspective · MoneyAfter building a test set, the team went from one evaluation a week to dozens a day. Once the test set existed, the accuracy of expert grading went up and up, and removing a lot of complexity helped — the team didn't need many of the added layers. Focusing on the right embedding model, the right chunking, and the right alpha in BM25 is the right thing to do, and accuracy increased over weeks.
AI Engineer · How to Test AI Agents When No Benchmark Exists — Nathanael Weill, Air Canada
Principle · MoneyAim for a script that runs the full evaluation and returns results in about five minutes. Exact-match and code-based gradings are very fast, while LLM-as-a-judge approaches are slow and noisy. LLM-as-a-judge produces so much noise that solving the judge problem is almost as hard as solving the original problem. Finding the right rubric (RNS) for an LLM-as-a-judge is a very hard task.
AI Engineer · How to Test AI Agents When No Benchmark Exists — Nathanael Weill, Air Canada
Principle · MoneyThe speakers emphasize not using these AI tools to just create content that is actually slop. The tools exist to automate tedious tasks (like captions and rotating objects that would take hours) so you can treat content creation like a craft, learn constantly, and improve. Content creation is your craft and your voice, and you always have to be in the driver's seat — otherwise nobody will watch. The goal of the tools is to get closer to the thing you want to say and your voice; your role is to think of the big ideas and treat AI as helping you execute on them.
Presented as the speakers' strongly held philosophy.
AI Engineer · The AI Content System Behind 200M Views — Sarah Chieng, Cerebras & Cova Lee, OpenAI
Principle · MoneyTo find success as a creator in the Instagram/YouTube ecosystem, one recommendation is to live in the real world, experience a lot of things, and pay attention to what draws you to them. Taste is an overused word, but everyone has a personal POV in how they live their lives and experience things. It is important not to just absorb things from the internet and adopt that mindset, because it prevents developing a strong and unique POV. The speaker deeply believes they and Sarah succeeded in part because of strong POVs cultivated from real experiences. Treat yourself as a curator of your life and experiences, and be intentional about what you choose to do — a good framework for content and anything you do.
Presented as the speaker's beliefs and recommendations.
AI Engineer · The AI Content System Behind 200M Views — Sarah Chieng, Cerebras & Cova Lee, OpenAI
Creator perspective · MoneyLLM judges have preferences: most models prefer themselves, and OpenAI models particularly prefer their own family. When using an LLM judge to find the best model for your task, be careful and use different model families to ensure fairness. Anthropic and OpenAI models appear to have post-training that makes them rate each other poorly.
The 'war of hating each other' claim is speculative ('it seems like').
AI Engineer · Why AI Writing Still Reads Like Slop — Louis-François Bouchard, Towards AI
Principle · MoneyIn spec-driven development, the spec is the constant source of truth for the product, analogous to how source code compiles into machine code.
NNgroup · Specs Are The New Source Code (feat. Thomas Watkins, NN/G Guest Instructor)
Creator perspective · MoneyThe guest argues that because AI is writing a lot of the software, it needs something to read and refer back to — on the design-intention side, you go into the actual widgets and areas that the code needs to listen to, so the tools can read the spec and refer back to it when building the software.
NNgroup · Specs Are The New Source Code (feat. Thomas Watkins, NN/G Guest Instructor)
Principle · MoneyAnything you don't articulate to the AI, it fills in with the average/defaults, producing a mediocre product.
Also discussed by Y Combinator (unclassified source — not counted until classified). Not counted as independent sources.
NNgroup · Specs Are The New Source Code (feat. Thomas Watkins, NN/G Guest Instructor)
Principle · MoneyThe guest endorses that anything you don't articulate to the AI, it will fill in with the average — it goes with its defaults. The purpose of specs is to force yourself to articulate the many things you want the design intent to impact in the reality of your code.
NNgroup · Specs Are The New Source Code (feat. Thomas Watkins, NN/G Guest Instructor)
Principle · MoneyMax Kanat-Alexander predicts that in the future, engineering rigor will be applied to the degree of disposability of the software. On the disposable end of the spectrum, none of our engineering rigor practices matter at all; on the durable end, all the practices built over decades matter greatly. Disposable and durable are not black and white; there is a spectrum, and what changes across it is how much effort you put into engineering rigor, both immediate effort and effort over time.
AI Engineer · Does Code Quality Still Matter? It Depends on This — Max Kanat-Alexander, Capital One
Principle · MoneyStandardization reduces the cost of rigor by eliminating possibilities. Without standardization, you have to set up rigor tooling every single time — you can still have the agent code up all your tools, but you have to do it every single time. Having every front-end or back-end framework in existence makes rigor more expensive.
none stated
AI Engineer · Does Code Quality Still Matter? It Depends on This — Max Kanat-Alexander, Capital One
Principle · MoneyThe most important thing to care about is reducing the cost of effort and the cost of rigor, not rigor itself. Use a different set of practices that decrease the cost of engineering rigor rather than decreasing rigor.
none stated
AI Engineer · Does Code Quality Still Matter? It Depends on This — Max Kanat-Alexander, Capital One
Principle · MoneyAdit Abraham states that if you're going to deploy document AI in production, you need not just great parsing accuracy — you need tools for evaluation and review, and for constructing workflows around the outputs.
The a16z Show · The Missing Infrastructure for AI Agents: Documents | Reducto’s Adit Abraham
Principle · MoneyAdit Abraham stresses that for high-stakes document processing, it is critical to know that the inputs you're dealing with are trustworthy, because everything builds on them. Overworked professionals tend to trust tool outputs without double-checking, so errors compound silently.
The a16z Show · The Missing Infrastructure for AI Agents: Documents | Reducto’s Adit Abraham