Hacker Newsnew | past | comments | ask | show | jobs | submit | raooll's commentslogin

1) Smartphones 2) Coding agents.


@nickyweek, please provide your contact email.


Safest would be the tech you already know well.


I have generally move away from JS/TS default. I even avoid react unless it is absolutely necessary. I am moved to more simpler choices when picking tech stacks. I go for the old & tested rather than the fancy, new & shiny choice.

This move has allowed me to maintain a project for long times while still maintaining my peace of mind.


I really like https://ai.pydantic.dev/ as it builds on existing developer knowledge and there is not much to learn in the framework itself.


I would go with https://tanstack.com/ as that has components to replace almost all Next.js features.


I'm planning on using tanstack start in my next project. Seems to have had a lot of good press recently


Hey Joshua,

I have been struggling with the same paradox, all AI moat seems to be to remove humans. Although that might be possible in a distant future I don't see it happening anytime soon.

Instead I want AI to be my assistant like and IDE, where it help me get things done faster that I can do it by myself but I am always in command. Also what if there is an AI assistant for each code repo, that understands that repo in its entirely and is downloaded when you clone the repo, it has deep knowledge on the code base, it scope and restrictions, how the code changed over time etc etc.

We can provide Small language models for existing codebases for large organizations like banks which run cobol or some other older tech. This AI then empowers there limited developers to punch way above their numbers and help them developer better, cleaner and faster.

This is one way I am thinking on AI/LLMS as they keep shrinking exponentially and keep becoming more powerful.

I think specialized single purpose embedded AI/LLMS will be the future.


Can I ask what if anything else you do besides work that interests you ?


Big fan of Trigger.dev here

We’ve been using it for a while and I’m a super happy customer. The scheduled task feature has been a lifesaver—super simple and just works exactly as expected. Also want to call out the Dev mode: it’s an absolute gem, makes testing and iterating so much smoother than anything else I’ve tried.

Really appreciate the thoughtfulness that’s gone into the product—excited to see the upcoming MicroVM + checkpointing work land!


This seems like a blatant AI comment


Hey, are we talking about deploying a LLM server here or LLM powered applications ?


LLM powered applications


I have deployed a few. One with a million registered users.


Cool! Would you mind if I interview you for 15 minutes? Here is my linkedin if you want to connect over there! www.linkedin.com/in/luis-liu-li


sent


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: