Juan Cruz Martinez

The tighter you try to control things, the less control you actually have. And as a consequence, everyone will be less happy and less productive.

It's confirmed, the credits are a consolation price for reducing your usage limits by 50%. SI seems more expensive than AI 😂

Codex remote connections are far superior than anything else I tried.

Ta-da! Is this because now my €200/month plan provides the same value as the €20/month was providing until yesterday? I need to make good use of them.

If every hard decision still ends with you, you haven’t built autonomy. You’ve built a queue.

Anyone out there really calling it out SI? 🤣🤣🤣

Life is life, work is works and AI is a tool newsletter.thelongcommit.com/p/when-ever...

Today I opened vim and I struggle to access some shortcuts. I thought it was already muzzle memory. This is the true sign that agentic coding is deteriorating my coding skills.

A move into engineering management can change what you need from technical reading. If your work has shifted from writing Rust to architecture decisions, a book on system design may serve you better than more Rust internals.

AI made it easier for me to keep working into evenings and weekends. I was shipping more, but also putting in more hours. Was I more productive because of AI or because the extra hours? newsletter.thelongcommit.com/p/everythin...

It happens to me every single time! I set up an amazing recurring prompt, I’m amazed by how well it works. Two days later I couldn’t care any less about it, and I hate the fact that I don’t like the output and I need to correct it all the time. My love for AI is short lived!

When you ask your manager for more responsibility, name the engineering work you want to own. Agree on timing and a follow-up so you can check whether your role has actually changed.

If an agent helps change its own access controls, who independently checks that small edits haven't weakened monitoring or tests? newsletter.thelongcommit.com/p/slowing-a...

An AI scan flags an essay from a writer you've enjoyed for years. Even an accurate score leaves the trust question open: is their judgment still worth following? My July essay: newsletter.thelongcommit.com/p/the-ai-sc...

The greatest productivity hack is not AI, it’s your engineering culture.

Code gets you started. Judgment, communication, and trust carry you further.

If you're considering another internal role, start coaching a colleague through the module only you know how to maintain. Making that responsibility transferable can keep it from tying you to your current role.

An engineer's “Cursor thinks you’re wrong” may be worth hearing. Ask what they recommend and what evidence supports it; using the model can remain part of the discussion. newsletter.thelongcommit.com/p/the-pr-sh...

By May, I was letting agents’ code through on non-critical projects. What bothered me was how much that category had grown: work I once considered critical had started feeling routine. newsletter.thelongcommit.com/p/i-didnt-k...

The real thread to your software engineering job is not AI. It’s the engineer who knows how to use AI properly, specify requirements clearly, evaluate architectural trade-offs, and can optimize for token consumption.

If you want to be an engineer you’ll have to spend a lot of time learning real engineering first. Once you are a capable engineer, start with the AI tools. The other way around doesn’t work.

The strongest engineers are not always the fastest coders. They are the ones who make the fewest expensive decisions.

AI researchers: AI will kill us all in 10 years AI CEOs: we need 300B so we kill us all before someone else’s AI kills us all. Me: how many R’s in strawberry?

You’ll learn more in one hour of doing something than in a lifetime of prompting about it.

The best engineering leaders don’t optimise for faster decisions. They optimise for decision quality. Speed only matters when the decision creates meaningful progress.

Do people saying learn AI today or you'll fall behind realize that whatever you think you learn about AI last year applies today?

A difficult conversation can run past its calendar slot. As a new engineering manager, leave some time unassigned by accepting less planned work, so a longer conversation doesn't take your evening. newsletter.thelongcommit.com/p/the-new-e...

Management filled my reading with leadership, DevRel and writing. After my holiday, I wanted to read something unrelated to work, too. newsletter.thelongcommit.com/p/when-ever...

Considering your first engineering management role? Look at the work you'd be learning: helping people deliver, communicating intent, handling competing priorities. If those responsibilities appeal to you, don’t let doubt take over. newsletter.thelongcommit.com/p/the-case-...

I finally tackled a monitoring dashboard I'd been postponing after AI gave me a starting point. It took about as long to finish, but I no longer dreaded the work. newsletter.thelongcommit.com/p/everythin...

When you ask a senior engineer to coach a colleague through a difficult decision, make room for it in their workload. That coaching needs time beyond the review comment that started the conversation.

You don’t need to chase every new tool to stay relevant. But you do need to keep learning before the market forces you to.