Creator perspective · MindHuberman describes his strength protocol: a moderate warm-up set prepares the nervous system for the heavier load, then two work sets spaced three to four minutes apart — all-out efforts in perfect form taken to momentary muscular failure, limited to two sets. He says people go badly wrong with warm-up approaches, though data and online advice are catching up to his method.
hedged
The a16z Show · Andrew Huberman’s Three Rules for Better Health and Focus
Principle · MindHuberman says resistance training is the really sharp blade of health improvement, more than cardio, and that in terms of body recomposition, mood, and brain health, you're not going to find a more potent stimulus. He still does cardio in addition to resistance training.
The a16z Show · Andrew Huberman’s Three Rules for Better Health and Focus
Principle · MoneyNikhil Harithas endorses building bookkeeping logic to handle dropped or unacknowledged sandbox creation requests — tracking whether a sandbox was started, its name, and how to route user requests to it — plus retry logic so the client and sandbox understand where they are in a given conversation. Failure cases include AWS acknowledging a session start request but never creating the sandbox, and mid-session requests dropped without an acknowledgment back.
hedged
AI Engineer · Your AI Agents Shouldn't Care Where They Run - Nikhil Harithas, Factory
Principle · MoneyNikhil Harithas endorses giving customers optionality to hydrate their compute layer however and wherever they want: saving sandbox snapshots to their own S3 or database of choice, hydrating snapshot compute with secrets from their own secret store, and bringing their own build and containers of choice. When building compute infrastructure, plan for customers to own their snapshots and secrets rather than locking them into your platform.
AI Engineer · Your AI Agents Shouldn't Care Where They Run - Nikhil Harithas, Factory
Principle · MoneyNikhil Harithas says the two questions that kept recurring in enterprise agent deployments are: who is the agent acting on behalf of, and where does it run?
AI Engineer · Your AI Agents Shouldn't Care Where They Run - Nikhil Harithas, Factory
Principle · MoneyNikhil Harithas endorses providing a spec to an AI agent (Droid) and asking it to generate a driver that fits the spec — the spec acts as its own test. He frames this as a very good form of test-driven and spec-driven development: if the implementation satisfies the given spec, it is basically ready to go.
AI Engineer · Your AI Agents Shouldn't Care Where They Run - Nikhil Harithas, Factory
Principle · MoneyNikhil Harithas argues that for agents, the compute plane needs to be spun out into its own thing, separate from the traditional data plane and control plane.
AI Engineer · Your AI Agents Shouldn't Care Where They Run - Nikhil Harithas, Factory
Creator perspective · MoneyKatie Givens credits Mark Kosiglo with teaching her that to understand any metric you need two things: context (the benchmark or goal) and trend (the direction over time). She illustrates with hypotheticals: ten outbound meetings booked last week is meaningless without context — with one rep it's excellent, with 30 reps against a goal of 90 it's a failure; and beating goal (10 vs 5 meetings) but trending down from 20-30 meetings over 12 weeks still signals a problem. A win rate dropping from 50% to 25% over 6-8 quarters is the kind of trend that prompts investigation into what changed.
30 Minutes to President's Club | No-Nonsense Sales · #622: Sales Enablement Playbook: How To Win Your First 90 Days
Creator perspective · MoneyRemote prompt execution is a technique where the initial prompt doesn't contain all of the prompts and instead instructs the agent to retrieve them from another location. Roey Ben Chaim explains it is legitimate but problematic because someone can take over the domain or change the prompt entirely at a later point in time.
AI Engineer · You Can't Tell What an Agent Skill Will Do by Reading It - Roey Ben Chaim, Zenity
Principle · MoneyDharmesh Shah recounts that when HubSpot was raising venture capital, VCs universally rejected the pitch because no marketing software company had reached billions in value — and HubSpot built one anyway. His endorsed heuristic: expect VC skepticism when your startup targets a category with no prior billion-dollar winner, and treat it as a signal of whitespace, not disqualification.
The a16z Show · HubSpot’s Co-Founder on Finding Your Next Startup Idea
Principle · MoneyThe speaker's stated principle: anything that can be solved with a structured query deterministically should be solved with a structured query deterministically — don't use inference to solve something a query can solve. He gives the example that a question like how many customers or how much revenue you're getting from Ireland should be a database query, not thrown at a 300 IQ model, even though the data is probably there somewhere in Slack messages and emails. He notes he has gone on record saying this, and the other speaker agrees it doesn't make any sense to do otherwise.
hedged
The a16z Show · HubSpot’s Co-Founder on Finding Your Next Startup Idea
Creator perspective · MoneyManaging CI is hard because the shape of the workload differs from what hyperscalers like AWS, GCP, and Azure are good at. CI jobs resemble desktop applications — like video games on a gaming PC — caring less about many threads and more about one single thread being really fast. The real way out is a cloud extremely good at running CI workloads.
Y Combinator · The $1B Company Fixing AI’s Software Bottleneck
Principle · MindGreene endorses the general principle that there is a way of experiencing things where you're inside them and not outside. He applies it to movies — he likes, especially with films from a different period or the past, to feel inside the actual movie itself — and notes movies are long-form compared to social media. Being immersed in how others are looking at you prevents experiencing the actual cultural moment, and culture can and should be sublime if experienced differently. Our best experiences, he says, come from being really present to work and experiences.
Andrew Huberman · Access the Sublime & Live Your Best Life | Robert Greene
Creator perspective · MindRobert Greene argues that the speed a person can access reflects the slowness of their underlying process — visible brilliance comes from slowed-down thinking. We only see the end product and assume skilled people fire on all 12 cylinders all the time, missing the slow process behind it. He cites John Elway, who said the secret to his success was that everything slowed down on the football field: through mastering a process, his brain could slow things down and see things others couldn't see. The generalization: mastering a process deeply allows perception to slow down so you can see what others cannot.
Andrew Huberman · Access the Sublime & Live Your Best Life | Robert Greene
Creator perspective · SocialIn a real relationship you get constant reassurance, but in a situationship one day you're the sexiest woman he's ever laid eyes on and the next you don't exist enough to warrant a text. This unpredictability triggers 'intermittent reinforcement,' which the host explains is like the feeling of hitting slot machines in Vegas. Because the reward (a kiss or a text) is unpredictable, your brain starts chasing it and releases a way bigger spike of dopamine when it finally arrives.
Call Her Daddy · The Situationship Survival Guide
Principle · MoneyIn a regulated industry, permission to act requires showing three things for every agent action: who the actor was, what policies allowed it, and the signing of the record. When audited, the questions will be who acted, who allowed it, and to show the record of what happened — not whether the model is smart enough.
AI Engineer · A Valid Key Let an AI Agent Delete Production - Sagar Batchu, Speakeasy
Creator perspective · MoneyThe crypto angle for agent payments is the ability for digital dollars to be sent cheap, fast, and permissionlessly anywhere in the world. Crypto has a particular role in microtransactions: an agent can spend one one-thousandth or even one one-millionth of a cent to access an API, a service, or a paywalled website seamlessly and trivially. Anyone running a business who wants their AI agent to accept a transaction can do so with crypto payments.
AI Engineer · Most of Your Docs Readers Are Now AI Agents - Hahnbee Lee, Mintlify & Lincoln Murr, Coinbase
Creator perspective · MoneyThe source describes low-mix, high-volume manufacturing as very hard to automate in the US because changing the line frequently is very expensive, and advises against investing in bespoke automation lines for low-volume production. Instead, the recommended mantra is 'design for manufacturing': build transferable, adjustable production lines so you can iterate prototypes faster and get components to rate production quicker, instead of fixing a design then rebuilding subsystems for scale. This is framed as a warning from Tesla and SpaceX.
The a16z Show · The Startups Rebuilding American Industry | The AD50
Principle · MoneyAn agent-ready design system is built from thoughtful component design, consistent naming conventions, and descriptive documentation, plus layered agent guidance such as examples, guidelines, and skills.
Figma · A step-by-step guide to an agent-ready design system
PrincipleJimmy Carr endorses evaluating beliefs by asking whether you believe something because you asserted it in the past and don't want to be wrong, or because it makes the most sense — and checking this when you find yourself defending a belief.
PowerfulJRE · Joe Rogan Experience #2571 - Jimmy Carr