Sasha Dzeletovic

[Garry Tan @garrytan]

The biggest alpha leak of 2026 is that you can tokenmax $10k/mo with OpenClaw/Hermes + GBrain and get the AI that everyone will have in 2028 for $100/mo, but you can get it now, and that is the biggest single unlock you can have vs your competition

[Sasha Dzeletovic @chordol]

In 2010, @Futurescale warned: "Don't hitch your dinghy to the Titanic."

https://t.co/yM2Bg10KNM

The lesson? Don't over-couple to platform-specific features. Flex died exactly as predicted. Teams burned down massive apps and rebuilt from scratch.

16 years later, same warning for AI build platforms.

Ask first: What are my portability options? What's the cost of leaving?

Platforms die. Portable architecture survives.

What excessively expensive over-coupling have you experienced?

[Sasha Dzeletovic @chordol]

After 25 years in software - from PHP and Flash to Node.js, Python, big data, Kubernetes, Terraform, microservices, and every JavaScript framework in between I've landed somewhere unexpected: Ruby on Rails. And I've never been more productive.

For years, I was constantly underwater. The JavaScript and Cloud worlds kept reinventing architecture. New patterns, new paradigms, new ways to solve problems that were already solved. By the time you learned one approach, three more had replaced it.

Then I gave Rails another shot.

In the last year, I shipped Kontekst. I'm finishing two other products. I rarely reach for Node.js or Python anymore - only when there's a specialized library, and even then, I just plug it into Rails. WebSockets, real-time features, everything I need, it's all there.

The joy of building is back.

Twenty-five years to learn this: the best technology is the one you forget you're using.

#RubyOnRails #SoftwareEngineering #DeveloperProductivity

[Sasha Dzeletovic @chordol]

Recently I built a complete software product with real-time communication and AI - entirely by myself, using AI. Before AI, this level of production readiness would have taken years or simply wouldn't have happened.

So what?

It means we can build businesses that aren't purely optimized for money. When you need funding and a team, money easily becomes the only common language. But building with AI, you can afford to preserve the art, the craft, the soul. You can afford expression that matters to you but has no place in a pitch deck. You can create value that's including of, and greater than money alone.

#AI #Startup

[Elon Musk @elonmusk]

Powerful words

[DHH @dhh]

We've switched all our dev setups to localhost. Not only can it have subdomains, but it's also a secure context by default, so you don't need any SSL cert to use all the features that depend on that. We just reserve a specific port per app. So good, so simple.

[Y Combinator @ycombinator]

At first, prompting seemed to be a temporary workaround for getting the most out of large language models. But over time, it's become critical to the way we interact with AI.

On the @LightconePod, Garry, Harj, Diana, and Jared break down what they've learned from working with hundreds of founders building with LLMs: why prompting still matters, where it breaks down, and how teams are making it more reliable in production.

They share real examples of prompts that failed, how companies are testing for quality, and what the best teams are doing to make LLM outputs useful and predictable.

0:58 - Parahelp’s prompt example
4:59 - Different types of prompts
6:51 - Metaprompting
7:58 - Using examples
12:10 - Some tricks for longer prompts
14:18 - Findings on evals
17:25 - Every founder has become a forward-deployed engineer (FDE)
23:18 - Vertical AI agents are closing big deals with the FDE model
26:13 - The personalities of the different LLMs
27:26 - Lessons from rubrics
29:47 - Kaizen and the art of communication

[pash @pashmerepat]

Boris from the Claude Code team explains why they ditched RAG for agentic discovery.

"It outperformed everything. By a lot"

[Sasha Dzeletovic @chordol]

@mattshumer_ OR... hear me out... Ruby on Rails 8. I'm using Claude Code to build Rails 8 project smoothly.

I started that approach because I am largely ditching the JS frontend jungle.

It seems to be working well with LLMs because it's one pattern for the entire project.

[Sasha Dzeletovic @chordol]

@Mikel_Vu Realized Next.js is overengineered and neither server nor browser friendly. It’s not a system I’d choose to maintain.

The buildup to that was years of wrangling React-to-server state and serialization.

[Sasha Dzeletovic @chordol]

@paulg @headinthebox I've been programming for over 25 years and I had to look up what SWE means.
We will all very much need people who understand systems to manifest these systems, for a very, very long time.

[Matt Shumer @mattshumer_]

<developer_note> is one of the most powerful, yet underutilized prompting techniques.

How (and why) to use it:

[Sasha Dzeletovic @chordol]

@p_ch @rails Twenty years later, I'm there and I'm happy.

[Sasha Dzeletovic @chordol]

@TurnerNovak These are also some of the best questions to ask in contrast to the silly made up scenarios.

[Sasha Dzeletovic @chordol]

@Madisonkanna Stick with vim and use Claude Code. Bonus if in split tmux session.

[Sasha Dzeletovic @chordol]

@ASpittel Preach! AI is really not changing that. It just further surfaces the importance of being able to articulate a coherent system. And that’s after actually being able to understand the business problem.

[Sasha Dzeletovic @chordol]

Here's a fun word: cvrkut. It's means chirp in Serbian. Pro tip: roll the "r" to get an extra vowel.