Migrating my entire frontend from MUI to Shadcn today. Feels like I’m purposely signing up for pain, but also… can’t avoid it forever. Perfect Sunday project, right?

Spent the last 2 months mostly ignoring one of our SaaS. We're at $1.8k MRR, just sort of coasting. Decided it's time to actually try-going to push for $5–10k/month. More marketing, more usability work. Let's see what happens when we actually do the work.

Migrating from Material UI to ShadCn sounded like a pain. But honestly? Having Claude generate code made it feel almost... trivial. UI went from "good enough" to actually polished, fast. Way more leverage than I expected.

You know it's been a good day when you've exceeded the monthly CC x20 plan's cost in token usage by 8 pm.

Some weeks you don’t ship features. You clean up technical debt, organize documentation, and fix the stuff nobody will notice. That’s this week. Not glamorous, but it pays off every time I revisit old code.

Is Claude’s Opus cap actually lower across the board or is it just the X20 plan getting these new limits? Never saw these warnings before.

Was two days deep into swapping Material UI for ShadCN on my React project-thought it would make everything look modern as hell. Still looks like shit. Turns out it’s not the library, it’s my design sense. Back to tweaking features. Ship > shiny UI for now.

Had my first real "wait, am I actually ok with this?" moment watching Claude run Terraform commands in my terminal. I know it’s my job to double-check, but damn-feels strange letting claude code handle infra.

Quota upgrades are underrated proof. It’s one thing to land customers. It’s another when they outgrow the plan in weeks. 100 → 200€ 200 → 400€ Usage-driven growth is the dream. Makes 10k MRR look less like a fantasy.

Churn rate: 0%. Not magic-just relentless focus on solving real problems fast. Example: When a major client asked for monday.com integration, we delivered it in a day. Customers stay when they feel truly heard. That’s the moat.

my favorite transformer fact: you can trace the meaning of a word as it gets updated by a crowd of vectors, each yanking its meaning based on who's shouting loudest in context reminds me of X tbh

List of stuff I wanted to do today: - Ship new feature - Add AI Assistant Actual: - Add missing error toast - Rewrite empty state text… again - Redo our Dashboard for the 100th time (i suck at design) All that to prepare for the next push.

People obsess over building fast. My real speed cheat code? Writing code markdown with ruthless detail before touching logic. Letting the autopilot (Claude Code) handle implementation after a rock-solid plan beats winging it in chat every time.

Was finally gearing up for the LiveKit phone agent launch… Then saw OpenAI just dropped their Realtime API with gpt-realtime support and SIP phone calling. Guess who's spending the rest of the day refactoring. Priorities keep shifting in real-time too.

Facts.

car work sessions used to wreck me motion sickness after 30 minutes, guaranteed but with claude code up, i find myself glancing out the window way more it’s not perfect, but now i survive the ride

days left before vacation: 5 projects left to finish: 3 energy left after wedding weekend: questionable sometimes you just have to eat the chaos and sprint

Building something new is the best kind of procrastination.

Wild how easy it is to lose track of wins when you’re deep in the build. Two months ago, we just launched the beta. Now? 100k products imported, 200k workflows run, managing 60+ Shopify stores. Did not expect this pace at all. Perspective is weird.

Did user calls (inspired by Joschua Sutee) and walked away with a to-do list longer than my original roadmap. Every product decision felt 5x easier after. Not wasted time at all.