Uber just capped each dev at $1.5k/m on Claude Code. Why? Subsidized tokens hook devs โ they get the company onboard โ it spreads org-wide โ SOC 2 forces them to full-price API โ bills explode.
Consumers get subsidized tokens. Enterprises pay full price. Well played Anthropic
"People actually pay for Buffer?"
Engineers, you should. Writing code stopped being the edge. AI write better code than you at 4am. What is left? Taste, system design, and an audience that trusts your voice. Virality picks you, you do not pick it. So post everywhere
Even LinkedIn (yes, cringe). it is sending me job offers + real users testing my product and where I got to meet people (for real). Use any tool. Buffer is free for 3 platforms.
Code got cheap. Being known is the game.
Unclear if a durable trend, but CEOs and CTOs are back to coding with a fury, thanks to coding agents.
I have public company CEOs sliding into my DMs (and โInMailโ) telling me about falling in love with shipping software again thanks to Claude Code and Vercel.
โDream accountsโ that we always wanted to work with, where in the past the C-suite would hardly understand the infrastructure until much later in the game.
Coding agents are the ultimate PLG-fication of the enterprise. Bad, legacy software canโt hide anymore. The stack that works is self-evident to the entire organization, from intern to CEO.
I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.
As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!
I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.
It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?
88ms => 1.5ms
150K allocs => ~500 allocs
Incredible right? Nope.
My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.
This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.
The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.
Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.
AI agents haha moments:
1. You prompt โ wait โ receive.
2. They run while you're not watching, and deliver value.
3. They constantly help you in the background.
Most of us are at stage 1, a few on stage 2. I'm building stage 3 with https://t.co/2TBWM4WiH3 ๐ฑ
Your company on auto-pilot? Let AI agents pick the right ticket while you sleep:
/loop overnight "solve every Jira/Linear ticket"
Small ticket + good design = it works.
Endgame: agents create tickets from reviews, support, analytics, and let them ship.
You're not in the loop.
Remember when copy/pasting 3 lines from StackOverflow without understanding them was considered reckless? We now ship entire codebases nobody has read.
Running my users agents with https://t.co/VcUfaYNoDa, but the user sandbox get totally stuck from time to time and had to restart the sandbox. Wish I could see sandbox states somehow. Having my eyes on cloudflare/vercel sandbox. anyone tried them?
Last week at Buffer, after 11 years.
4-day weeks. Holidays. A 3-month sabbatical. None of it really worked, my brain was always on, one eye on this calendar.
Strange satisfaction removing it.