Lithos.eth

it is only april and over seven hundred million dollars has already vanished forty seven plus protocols have been completely wrecked so far in 2026 this shows exactly how violent the timeline is right now hacks are not getting luckier they are getting smarter bad actors are now using ai to scan massive codebases and find the tiny flaws before we even launch the classic defense of using human auditors to fight machine speed is basically is not enough we cannot keep deploying naked smart contracts and hoping for the best the only way to survive is moving execution into secure enclaves and using cryptography to hide the actual plumbing from the scanners if we do not build systems that fight back we are just exit liquidity for the builders how are you upgrading your infrastructure to fight off these automated agents ? and for the everyday users how are you protecting your own stash when every single protocol feels like a target

Jonah's filters work because they're measuring the one thing AI can't commoditize The internet made knowledge worthless almost overnight. Google turned recall from a skill into a commodity, and most people barely noticed it happening. AI is doing the same thing to intelligence. Being able to reason through a hard problem used to be a real competitive edge, and that gap is closing faster than most people want to admit. What holds through both shifts is EQ. Not the vague soft skills version of it, but whether people feel heard when they talk to you, whether they'd call you when something goes wrong, whether they'd take a meeting just to stay in your orbit. That's what these two filters are actually screening for. Response behavior tells you how someone handles a stranger asking for their time, which is the same muscle as a deadline or a hard conversation. The money question tells you if they can sit with failure in front of someone who matters. Both are EQ tests in disguise. As AI handles more of what's on a job description, the only thing that can't be abstracted away is whether people want to be in a room with you when things get hard. That's the skill worth building.

ended the month with 11,850k followers. the goal was 12k. see you next month 🏌️‍♂️

3 things most teams treat as launch-day jobs that should've been done 8 weeks earlier: 1. Defining the narrative you cannot write a campaign brief if you don't have a clear story. most teams discover this under time pressure which is usually the worst time to find it. 2. Identifying the first 100 believers. more emphasis on the term 'believers', not users. these are people who understand the vision and would tell others without being asked if you can't name your beleivers before launch, you don't have them yet. 3. Making the founder visible. launching a product with an anonymous founder is like opening a restaurant where nobody knows the chef. maybe it works, but you've successfully removed the most powerful trust signal you have. > unpopular opinion: most products are already failing by the time they launch bc the structure was wrong before the campaign *stops typing to catch a breather*