Pinned Tweet
I post about building SaaS, AI tools, and automation; the real timelines, what worked, and what didn’t.
I also share stuff I enjoy, random opinions on business and politics, and whatever else is on my mind.
If you’re building something or just want to talk shop, drop what you’re working on below.
Happy to connect.
187 followers.
200+ posts.
Still small, but I’m starting to see which posts get real replies and which ones die.
Figuring out what resonates with people building SaaS, AI, and developer tools.
If that’s you, drop what you’re working on.
Building in public only works
if someone is actually watching.
Otherwise you’re just journaling in public.
The algorithm doesn’t hate small accounts.
It just doesn’t care about them. (Including me)
Big difference.
Noticed I’ve been saying “I’ll do it later” to small things all week.
The later never really came. Just ended up with a longer list and less energy.
Starting to treat the 5-minute tasks like they actually matter. Annoying how well it works.
Small thing that helped this week:
I started writing the “done” criteria before writing any code.
Not a full spec.
Just 3–4 bullet points of what “this is finished” actually means.
Cut a surprising amount of scope creep.
Trump just posted the tracksuit version 😂😂
Readers added context they thought people might want to know
Donald Trump did not post the video; it is a satirical AI/CGI-generated meme with a fake @realDonaldTrump account header overlaid. No matching post exists from his accounts.
x.com/grok/status/21…
x.com/grok/status/21…
x.com/realDonaldTrum…
100 followers.
100 posts.
Not a big number, but enough to start seeing which posts actually get replies and which ones just sit there.
Still figuring out what resonates with people building SaaS, AI, and developer tools.
Drop what you’re working on if you’re in a similar space.
What’s one “best practice” you used to follow that you now mostly ignore on side projects or early-stage features?
Tried the approach from my last post on 3 real features this week.
Average time to ship dropped from 2.5 days → 11 hours.
Biggest win: stopped over-abstracting early. Wrote the boring version first, then extracted only what actually repeated.
The pattern that worked best:
Ship the narrow version
Measure real usage for 48h
Only then refactor
Curious — has anyone else noticed that “clean architecture” early on actually slows the first ship for most CRUD-style features?