Me pasan cosas con la comida

timezones are the best. you wake up, check Slack, and learn you’ve been unblocked for 6 hours.

The room went silent. We all stared. He wasn’t wrong. He went home yesterday and hasn't shown up today, but he wasn’t wrong.

We were debating which cloud provider has the fastest response times. Someone said google. Someone else said aws. One guy pulled up a latency chart. Another started talking about edge caching. then the junior dev whispered, almost apologetically: “localhost”

And that’s fine. The trade-off is what gave us the velocity to build, deploy, and scale faster than ever. But sometimes I wonder how many of us even know where our code runs anymore.

You can build the cleanest pipelines, the most redundant autoscaling, the prettiest infra dashboards... and one config file in Northern Virginia can still end your day. Every outage is a reminder that we’re guests in someone else’s computer.

I saw a dev refactoring today. Tab 1 ChatGPT. Tab 2 Claude. Tab 3 Gemini. Tab 4 Grok. Tab 5 Llama. He gave them all the same function. Each added different bugs. He merged them into one super-bug. It compiled. Why is no one else doing this???

Yesterday was my first day at AWS. *Not really But if you were affected, you should definitely post what this outage taught you about b2b saas.

*Sometimes differentiation is just semantics

If you’re still using “Loading…” in your product, you’re doing it wrong. Change it to “Thinking…” and you’re now an agentic AI startup Change it to “Dreaming…” and you’re pre-revenue but visionary Change it to “Becoming…” and you’ve raised a Series A

Then it fixes your fix. Then you both apologize to each other. By the end, you’re emotionally bonded to a stochastic parrot. And it still doesn’t run.

Coding with AI is fascinating. You start with a blank file and describe your vision. It generates code that’s almost perfect. Then you fix it.

Junior: writes code Mid-level: fixes code Senior: deletes code Staff: deletes the team that wrote the code Principal: writes a memo explaining why no code is the most scalable solution CTO: buys software to replace the code Then hires a junior to integrate it Circle complete.

Which caused him to spill a whole Mate on his MacBook. Which bricked the MacBook, delaying the board meeting. All from one rm -rf config.json. That’s systems thinking.

People say good engineers think in systems. They see how one line of code ripples through an entire architecture. I once saw a dev delete a single config file. It crashed staging. Which crashed prod. Which triggered PagerDuty. Which woke the CTO.

AI won’t replace engineers. AI is an engineer. It lies with confidence. It introduces breaking changes on Friday. It opens a pull request that fixes one bug and adds three. The singularity has already happened.

You joined @IFTTT on this day in 2011 🤯

The problem isn’t if they’re working. The problem is if they’re too productive. Because then you wake up and see they containerized everything. And renamed all your variables into Spanish.

If they’re pushing to main at 3am, they’re working. If they’re on a Zoom call with a rooster crowing in the background, they’re working. If they’re in Argentina, it’s summer, and you see palm trees reflecting in their glasses -- they’re working harder than you.

Remote work skeptics always ask: “But how do you know they’re working?” Simple 👇️

“Wisdom isn’t just seeing the answer first — it’s having the patience to let others arrive there on their own.” - ChatGPT