Daniel - Js Craft

On Success, Luck, and Values This is a less technical piece than what I usually write on this blog. Hope you enjoy it! Lately, I’ve been thinking a lot about long-term games. There’s a considerable amount of luck and timing that ...

Success isn’t just about hard work - it’s also about timing, luck, and the people you choose to share the journey with. I wrote this small rant on why the daily process, values and companions matter more than outcomes: www.js-craft.io/blog/on-succ...

Building a timeless foundation in AI Engineering & MCP in JavaScript I've started working on the MPC in JavaScript book. The more I dig into it, the more I find myself reflecting on timeless AI engineering tech - the kind of foundational knowledge that pays dividends f...

I have just published the September edition of the JS Craft newsletter - Building a timeless foundation in AI Engineering & MCP in JavaScript

👉 www.js-craft.io/blog/timeles...

How to get an "unfair" advantage vs AI? Build real human connections.
I've tracked that the a podcast episode takes me 8-10 hours to produce. Why do I do it? Well, learnings. But the most important is the motivation it gives me!
Interacting with someone who I admire === energy++

How to get an "unfair" advantage vs AI? Build real human connections.
I've tracked that the a podcast episode takes me 8-10 hours to produce. Why do I do it? Well, learnings. But the most important is the motivation it gives me!
Interacting with someone who I admire === energy++

Learned this the hard way! In dev mode, start by testing your prompts with smaller AI models. They fail quickly and often, which is exactly what you want. Larger models can mask mistakes, making debugging much harder. On top of that, smaller models are cheaper leading to a faster feedback loop.

Today I did a talk on MCP in front on 30+ devs.
I love writing and hate speaking. So much energy goes into how you speak instead of what you say. When writing, you can edit and refine. When speaking, it’s all real-time.
But being good at speaking will always gives you a huge edge!

How to make your side gig the most efficient as possible:
1. Have a baby
2. get a 9-5 job
3. Fit the side gig in the time left.
In the end, constraint inspires creativity. It forces you to get real very fast.

Thoughts LLMs && Coding
1. English has a more complex syntax & vocabulary than JavaScript. If an LLM can write decent English, there is a high chance it is able to write JS as well.
2. With a code interpreter, an LLM can validate its own responses & regenerate. This is unique & essential.

Will MCP become the HTTP of AI?

👥 I think the only truly long lasting moat is doing things that don’t scale. Showing up in person. Building genuine human connections one by one.

YouTube video by Kleiner Perkins How I Built A $30M Business Without A VC | David Heinemeier Hansson

C. Newport and DHH are right! I often find myself having 2 different discussions in MS Teams while attending a meeting on a 3rd topic. In the modern workplace, we’re battling to find space for deep work and flow. And we need them, not just to be productive, but also to be happy!
youtu.be/uAFCvQtjZ7o

💡 LLMs, with their natural language processing capabilities, will become the UI.
For example, you won’t need to learn Excel’s interface and functions. The embedded LLM will understand commands like: 'Give me the average of the top 50% of values from the salary column'.

Very few things give me better vibes than receiving reviews like this one from readers of my LangGraph js book: jscraft.gumroad.com/l/zbbgz

💰Where can I invest my learning today and earn dividends for years to come?
Think about how rewarding it would have been to deeply learn SQL 20 years ago, or to master CSS. You could have built an entire career on those foundational skills.
With the rise of AI , this question is even more important.

YouTube video by Js-Craft Academy 🎙 Nicholas C. Zakas -creator of ESLint- JavaScript and AI, writing well in the age of LLMs and more

In case you missed it, I’ve published really nice interview with @humanwhocodes.com, the creator of ESLint.
We discussed:
- AI engineering with JavaScript
- MCP and his new project, CrossPost
- building a long-lasting career as a software dev
And much more 👇
youtu.be/gqOgRt9XuuU

YouTube video by Nikhil Kamath Clips Vinod Khosla’s Brutal Career Advice For Every 20 Year Old

This is brilliant. Learn how to learn and adapt - knowledge compounds just like money.

www.youtube.com/watch?v=eE6e...

More #JavaScript AI frameworks please! Congrats, Sam and the team! Mastra AI is now at 100k NPM downloads!

💬 Python is great for training AI models, but JavaScript is the true language of web builders. Nothing beats the JavaScript/TypeScript ecosystem when it comes to shipping apps to production.

That's why I'm quite bullish on using JavaScript for AI Engineering.

Speaking of local models and Ollama. All of this real-time visual recognition is now possible with a local LLM (Llama) running on an M3 MacBook.
👉 x.com/i/status/192...

Nothing beats sitting on the terrace with a good coffee and some deep learning on a chill Saturday morning.

On that note, if you're looking for a solid read, my LangChain & LangGraph books are 30% off as a bundle!

👉 jscraft.gumroad.com/l/zbbgz

YouTube video by Js-Craft Academy Using LangChain and LangGraph with local models

Been getting emails about using LangGraph AI agents with local LLMs for privacy. Thinking of adding a chapter to the book on this. Until then, here’s a video on installing Ollama, pulling Llama 3.2 locally, and integrating it with LangChain.js:

youtu.be/pKK07kObVnw?...

More and more, I see files on GitHub that mix human language and code. I'm starting to wonder if Andrej Karpathy might be right when he says that the most popular programming language will soon be English.

TIL: If you replace "https://github.com/repo-url-here" with "https://gitingest.com/repo-url-here", you can instantly turn any GitHub repository into a text digest of the codebase, perfect for feeding into an LLM.

The 6 hour podcast with DHH and Lex Fridman was too long for me to sit through, but it's full of great ideas on AI, learning, and programming. I didn’t want to miss out, so I took the YouTube transcript, passed it through an LLM, and ended up with this nice 170-page book for the weekend.

🎰 One of my bets is that, soon, most applications will have an integrated AI/LLM layer, just like they have a database. The AI layer could be used for stuff such as NLP capabilities, to enhance the UX, make sense of unstructured data, and even communicate with other AI's.

Super excited to share my interview with @humanwhocodes.com !
We talked about:
- AI assisted development
- MCP
- JavaScript vs. Python for AI integration
- Building a long-lasting career as a software dev
- The importance of writing well in the age of LLMs
👉 js-craft.io/blog/nichola...

NPM downloads for LangChain & LangGraph:
- LangChain.js up +123% YTD
- LangGraph.js up +311% YTD
Technologies are like stocks. Add the right ones to your portfolio, and some will become 10x baggers.

Both LangChain & LangGraph are covered in my book bundle, now 30% off 👉 jscraft.gumroad.com/l/zbbgz

What I love about books is that we actually pay full attention when reading. No multitasking, just focus. That's pretty rare these days, which is why I think books will remain - maybe even grow - as a learning environment.

GitHub - humanwhocodes/crosspost: A JavaScript utility for posting across multiple social networks at once A JavaScript utility for posting across multiple social networks at once - humanwhocodes/crosspost

@humanwhocodes.com 's Crosspost project is a fun and great example of a real JavaScript MCP server in the real world. Source code here: github.com/humanwhocode...

💬 LLMs will become the UI given their natural language processing capabilities. For example, you won’t need to learn Excel’s complex interface or functions, as the LLM will understand commands like: 'Give me the average of the top 50% of values from the salary column'.