Principle · MindOne of the strongest positions to apologize from is when you've already started making reparations. The coach tracked down the man who leaked the farm's location before apologizing; likewise, instead of apologizing and then changing your behavior, just start washing the dishes — then the fix is already underway when you apologize. Actions speak louder than words.
Charisma on Command · 6 Mindsets That Make People Respect You During Conflict
Report · MoneyMyth number two is the claim that companies that don't raise venture are going to get crushed by the companies that do. The speaker was scared out of his mind at every board meeting during the years at Moz when they couldn't raise their second round, fearing some other company would raise $10–20 million and Moz would be unable to compete. But the two companies that ended up beating Moz did not raise venture: SEMrush didn't raise venture until they were over $60 million in revenue (they had already beaten Moz by the time they raised their first round, and later went public), and Ahrefs still has never taken venture and is significantly larger than Moz. Additionally, 43% of companies that have IPO'd since venture funding started were venture-backed, meaning 57% were not.
MicroConf · Venture Capital Myths: Why 77% of Funded Startups Get Written Off
Report · MoneyAggregated statistics on venture-backed startups: using data from 1,120 companies funded 2008–2010, 77% were written off as a loss; the actual math shows a 90% failure rate for startups over five years; unicorn status occurs for only 0.9% of venture-funded companies (about a 1-in-100 shot); and 20% of venture-funded companies are acquired for less than 3x.
MicroConf · Venture Capital Myths: Why 77% of Funded Startups Get Written Off
Principle · MoneyThe founders advise targeting rent at about 10% of revenue when evaluating a retail lease (e.g., $200k rent implies aiming for at least $2M revenue), noting premium locations like Chadstone can push rent to ~20% of revenue while non-premium locations sit closer to 10%.
possible_numeric_error
Foundr · How Two Aussies Took On Samsonite & Rimowa - and WON ($100M+)
Principle · MoneyThe speaker explains that once you get the domain and own the IP for the category, you get the rest. He found an asset broker that helped him get 'founder' for Instagram, Facebook, and Twitter, brokering individual handle acquisitions for a couple grand each in the early days. A premium domain and matching handles build trust for a brand — would you trust a brand called 'july_luggage_official_69'? Probably not. July actually got the social handle before the domain, leveraging Meta's professional management program and existing professional relationships.
Foundr · How Two Aussies Took On Samsonite & Rimowa - and WON ($100M+)
Principle · MoneyThe founders break down retail unit economics: bulk shipping to a storage facility lowers shipping costs, and retail acquisition cost is calculated as rent plus variable wages plus CapEx on the fit out (with the landlord covering about half the fit-out CapEx).
resolved
Foundr · How Two Aussies Took On Samsonite & Rimowa - and WON ($100M+)
Report · MoneyAI agents have no government ID and cannot go through KYC, so they use self-custodial wallets and crypto rails; Coinbase will provide self-custodial wallets to any AI agent that wants to hold a stablecoin balance, raise money, or issue a token.
No Priors: Artificial Intelligence | Technology | Startups · Coinbase’s Everything Exchange: Agentic Finance, Stablecoins, and Tokenization with CEO Brian Armstrong
Principle · MoneyArmstrong argues AI agents deserve financial services of their own: they should have their own spend accounts, like a personal assistant or employee, and should not be unbanked.
No Priors: Artificial Intelligence | Technology | Startups · Coinbase’s Everything Exchange: Agentic Finance, Stablecoins, and Tokenization with CEO Brian Armstrong
Creator perspective · MoneyArmstrong argues frontier models are trained to be good generalists, but smaller open-weight models fine-tuned or trained via reinforcement learning on a company's proprietary data can outperform frontier models on a specialized task — for example, a small open-weight model trained on 100,000 compliance cases within Coinbase's proprietary data can outperform a frontier model on compliance review.
No Priors: Artificial Intelligence | Technology | Startups · Coinbase’s Everything Exchange: Agentic Finance, Stablecoins, and Tokenization with CEO Brian Armstrong
Report · MoneyCoinbase announced a tokenized stocks product, currently only available outside the US, with the SEC engaged for a US path. Armstrong predicts tokenization will extend down the list to private credit, treasuries, and bank deposits — everything will get tokenized, giving financial services modern rails. The host frames tokenization as broadening and modernizing the financial system, including bonds and assets that previously had much more friction to purchase, opening them to the world.
moderate
No Priors: Artificial Intelligence | Technology | Startups · Coinbase’s Everything Exchange: Agentic Finance, Stablecoins, and Tokenization with CEO Brian Armstrong
Creator perspective · MoneyWhen talking to founders of failed startups in your space, focus on learning why they failed, what they thought went well, and which channels worked — not whether people will pay, since people are already paying for products in that space.
Startups For the Rest of Us · Episode 852 | Getting Big Name Customer Case Studies and More Listener Questions (with Ruben Gamez)
Report · MoneyOnce your SaaS has product-market fit and is growing (~$20-30k/month), find ex-employees of competitors — people who no longer work there — and interview them. Rob says this is still advice he gives.
Startups For the Rest of Us · Episode 852 | Getting Big Name Customer Case Studies and More Listener Questions (with Ruben Gamez)
Creator perspective · MoneyKent C. Dodds says 'how many features are too many for an MVP?' is the wrong question; the real question is what is the minimal thing you can do to demonstrate that you solve a problem for your target audience.
AI Engineer · Build the Right Thing: Product Engineering (Part 1) — Kent C. Dodds, EpicProduct.engineer
Principle · MoneyQuoting Don Norman: don't ask somebody what the problem is, because they'll tell you the symptoms — and often they'll tell you solutions as well, and their solutions will be just as ill-formed as your own, maybe more so. Instead of asking generally 'what is the problem?', ask about specific behavior: 'What happened? Who was involved? What made it annoying enough to remember?'
AI Engineer · Build the Right Thing: Product Engineering (Part 1) — Kent C. Dodds, EpicProduct.engineer
Report · MoneyJim Clark explains that in the cross-app access scenario, the harness, sandbox, and gateway can define agent identity — who the agent is behaving on behalf of. By exchanging identity claims with your existing identity provider, you get back an ID-JAG (an authorization grant) that is part of the latest MCP spec, and with that grant the agent is granted access to the same authorization servers the company has been using.
AI Engineer · How Many Credentials Should Your AI Agent Have? Zero. — Jim Clark, Docker
Principle · MoneyKent C. Dodds observes that users often come to you with a solution — 'I have this problem, and here's what would solve it' — and some skip the problem entirely and just demand a solution, like 'I need an export to PDF button right here.' You need to be the one to ask questions and get at the heart of what the user is trying to do. When validating a pain point with other customers, don't ask 'is this a problem for you?' — instead ask 'when was the last time you experienced this pain and what did you do instead?'
AI Engineer · Build the Right Thing: Product Engineering (Part 2) — Kent C. Dodds, EpicProduct.engineer
Principle · MoneyKent C. Dodds argues the user is not always right — one user caring a lot about a particular thing is often not representative of other users — though users are often directionally right even when their specific ask is wrong. To sort out whether a complaint is representative, talk to other users, asking 'when was the last time you had this problem? And what did you do about it?' — Mom Test it across your other users.
AI Engineer · Build the Right Thing: Product Engineering (Part 2) — Kent C. Dodds, EpicProduct.engineer
Report · MoneyThe talk defines several eval types: a correctness eval asks whether the agent answered the question accurately; a faithfulness eval asks whether the agent stuck to the source documents without hallucinating; a capability eval asks whether your agent can do a task at all and is expected to mostly fail with a very low score because it gives you a hill to climb; a regression eval is one you expect your agent to pass completely or nearly completely every single time.
AI Engineer · From Vibes to Production: Evaluating and Shipping AI Agents That Work 101 — Laurie Voss, Arize AI
Principle · MoneyIf an agent hallucinates a stock price, that is a hard fail — don't ship on that. A financial analyst agent should produce actionable recommendations — buy, sell, or hold — not just summarize data.
AI Engineer · From Vibes to Production: Evaluating and Shipping AI Agents That Work 101 — Laurie Voss, Arize AI
Principle · MoneyUse a simple string check instead of an LLM judge when you're looking for a specific string somewhere in a report, such as whether a stock ticker was mentioned — an LLM judge is overkill for that.
AI Engineer · From Vibes to Production: Evaluating and Shipping AI Agents That Work 101 — Laurie Voss, Arize AI