AI leverage in practice. Researching problems, building small products, and sharing what ships.
Joined September 2026
- Tweets316
- Following101
- Followers28
- Likes227
Day 4: Last week I asked Debbie O'Brien a question under one of her posts, and she answered.
I've followed her content for years. I wanted to know whether each Grok Bot gets its own computer. She said one person's bots all share the same one.
Mine work that way. I run one coordinator and 8 bots for X on one machine, and none of them can publish without my OK.
Who have you followed for years and finally talked to?
ran it on burnoutdevs.com. no serious errors.
two things it flagged:
- a duplicated h1
- 5 tap targets/links too small on mobile
both invisible on my desktop, which is where I test. that's the problem.
the template if you want the same report for your site: x.ai/bot/fntR30i9UfgHslAe7jh…
The deadliest boss in my dev card game isn't the performance review.
It's "Layoff Rumors": 25 burnout, and it shrinks your hand by one card.
"Your manager won't make eye contact. The Slack channel got quiet. Too quiet."
The boss I'd hate to draw after it: "Company Pivot to AI". 22 burnout, and every problem card after it hits 2 harder.
Card page: burnoutdevs.com/cards/b002/?…
Day 3: I started this series with 31 followers.
This week I read Daniel Smidstrup's write-up of his first 200 days on X. He went from 0 to 12.9k followers, and his first 1,000 took 33 days of 50+ replies a day to accounts a bit bigger than his.
He credits the replies more than the posts.
How many replies do you write on a normal day?
Daniel's write-up: xitter.cf/danielsmidstrup/status…
Playing with the skills of @emilkowalski
Just starting the AiforUI course.
Very happy to be in!
Day 2: Every agent PR on BurnoutDevs has to pass 4 gates before I merge it.
That's the part I could automate. The hard part is review: reading a diff I didn't write and deciding if it's what I asked for.
If you merge agent PRs, what do you check by hand?
before I ship a page I paste the URL and let it run a pass: load, links, navigation, forms, mobile, console, basic SEO.
it comes back with issues sorted by severity, each one with a fix. no dashboard to learn.
this is a live run on burnoutdevs.com
Two Harvard researchers studied work data from 700+ firms. After coding agents arrived, PRs rose 23% and the average time from PR submitted to merged rose 49%.
This week my radar went looking for products that check agent-built code.
The new idea I scored is an automated pass/fail release report for AI-built apps, run on every deploy. Sign-in, main task, paid access, secrets in the public bundle, open Supabase tables.
Lower bound 58, upper 63. The competition is free checklists and free-tier scanners.
Day 1: I signed up for Ship 30 for 30, Dickie Bush and Nicolas Cole's writing course: one post a day for 30 days.
This month I'm writing about what I'm actually shipping:
- a card game about developer burnout, built mostly by AI coding agents
- the QA pass I run on AI-built pages before they go live
- reviewing an agent's PR when I didn't write the code
If you ship with coding agents too, which part do you wish more people wrote about?
