copynpaste, 7 weeks in - 6,300 people in the last 90 days - 4 paying users - 12 integrations shipped, every one from a user email - $0 on marketing built it because pasting between apps kept wrecking my formatting. turns out it wasn't just me

friday deploys are fine until you feel confident about them

good error messages are the cheapest devrel you'll ever ship

build the thing that annoys you first. you already know it's a real problem

a flaky test is worse than no test. it teaches the team to ignore red

what's shipping this weekend, even if it's tiny

what's one thing you wish your favorite dev tool would just delete

Compliments are nice, but hearing "recommended you because you're amazing at what you do." just hits different.

senior engineer energy is saying 'it depends' and then actually explaining what it depends on

when someone asks a question twice, write the answer where the third person can find it

nobody gets excited about client diversity but it's the thing i'd protect first in ethereum

do you read the whole diff of an AI PR or just the tests

one day you'll miss the version of you who was figuring everything out for the first time. keep building. keep asking questions. you're probably making more progress than you think.

sometimes the best debugging tool is explaining the bug out loud to someone who has absolutely no idea what you're talking about.

privacy shouldn't be a premium feature. you shouldn't have to choose between using a digital service and exposing your entire history to it.

would you give an agent a wallet with a $50 limit. what's the number where you say no

a wallet shouldn't just ask "do you want to sign this?" it should help answer "what will actually happen if you do?" signing is consent. understanding the outcome is the point.

if users need to understand gas, bridges, seed phrases, and 4 different networks just to use your app, the UX isn't finished yet.

the beautiful thing about zero-knowledge proofs: you can prove a statement is true without revealing the information that makes it true. still one of my favourite ideas in computer science.

a security feature nobody understands is a security feature nobody will use correctly. usability and security are not enemies. good engineering needs both.

an mcp server takes a weekend to build and a lot longer for anyone to trust