Joel Jacob Stephen

Shipping code & chasing curiosities.
Full-stack dev, AI rabbit hole survivor, Homelab enthusiast, Linux explorer.

[Joel Jacob Stephen @joeljstephen]

This is absolutely insane!! Two completely different game systems fused together and it’s somehow working seamlessly. My mind is blown.

[MotionViz @Motion_Viz]

you all, mind telling me what spiderman is doing on batman?

[Joel Jacob Stephen @joeljstephen]

Migration went surprisingly smoothly. I pointed Codex at my Fedora setup, had it translate everything into NixOS configs, and got almost the same system rebuilt with very little friction. My NixOS journey begins.

[Joel Jacob Stephen @joeljstephen]

Going to give NixOS a real shot and move over from Fedora. Its declarative nature feels perfect for AI agents. I can describe what I want, let an agent modify the system, and still keep the whole setup reproducible so I can rebuild it anytime.

[Joel Jacob Stephen @joeljstephen]

Going to give NixOS a real shot and move over from Fedora. Its declarative nature feels perfect for AI agents. I can describe what I want, let an agent modify the system, and still keep the whole setup reproducible so I can rebuild it anytime.

[Joel Jacob Stephen @joeljstephen]

Perfect Timing!!

[Cursor @cursor_ai]

Origin, our code hosting platform, is now live.

It's fast, easy to use, and deeply integrated with Cursor.

Get started by syncing your repos from GitHub.

[Joel Jacob Stephen @joeljstephen]

I wonder what this means for the AI landscape.

If Siri handles the everyday things most people use AI apps for today, many may not open a separate one as often. That could quietly shift habits and usage away from those platforms.

[Joel Jacob Stephen @joeljstephen]

The most impressive part of #WWDC to me is how Apple is integrating Siri deeply into daily apps, making use of our personal context but with privacy in mind.

AI becomes so much more useful when it understands my life but never felt comfortable giving that context to other AI apps. So this feels promising.

[Joel Jacob Stephen @joeljstephen]

The most impressive part of #WWDC to me is how Apple is integrating Siri deeply into daily apps, making use of our personal context but with privacy in mind.

AI becomes so much more useful when it understands my life but never felt comfortable giving that context to other AI apps. So this feels promising.

[Joel Jacob Stephen @joeljstephen]

Woke up to God of War Laufey and Marvel’s Wolverine game announcements.

Yeah, this is already turning out to be a great day!!

[Joel Jacob Stephen @joeljstephen]

Thank you @Google!!

[Mishaal Rahman @MishaalRahman]

💬 A more natural way to text

Yeah, so, um, the way we talk isn't the way, ah, we want to write...

Rambler in Gboard takes your messy, stream-of-consciousness speech & fits the important bits together into a concise message. Just speak naturally & let Gemini handle the rest!

[Joel Jacob Stephen @joeljstephen]

I really want smart dictation as a native keyboard feature on iOS and Android.

Not just transcribing what I say, but cleaning it up into clear notes or bullet points.

This feels like the kind of AI feature that should just be built in, instead of another app or subscription.

[Joel Jacob Stephen @joeljstephen]

March felt different.

The biggest AI launches weren’t just better models. They pushed AI deeper into the real pipeline of work.

Karpathy’s AutoResearch, orchestration in Cursor and Linear, and NVIDIA DLSS 5 all pointed to the same shift.

My March AI Roundup: https://t.co/rqinZ58pwL

[Joel Jacob Stephen @joeljstephen]

Just tried the new Glass UI and it’s pretty cool. Feels like a glimpse of where editors are headed, with the core experience built much more around AI and agents than the traditional IDE workflow.

[am.will @LLMJunky]

Nice. @cursor_ai just dropped their new "Glass" alpha, and they're leaning heavily into the simplified coding GUI trend that's been blowing up lately.

First impressions are really positive.

And just look at how insanely fast Composer 2 is. First impressions? Drop yours 👇

[Joel Jacob Stephen @joeljstephen]

We went from outsourcing to AI to being outsourced by AI.

[Polymarket @Polymarket]

BREAKING: New startup "RentAHuman" allows AI agents to rent humans to perform tasks they cannot physically perform themselves.

[Joel Jacob Stephen @joeljstephen]

@AskBaluBabu @residencyBLR @theresidency This is amazing, congrats!! 🔥

[Joel Jacob Stephen @joeljstephen]

I was scrolling through Instagram reels and am genuinely impressed by Meta AI’s translation feature. Seeing videos in different languages turn into English with matching voice and lip sync feels wild. Kind of scary how good this is getting.

[Joel Jacob Stephen @joeljstephen]

Running genuinely good models on consumer hardware is a real win.

This is what I want for OpenClaw style agents. Let them work with my calendar, notes, email, and files without sending over my data to a cloud LLM. Still early, but I’m hopeful we’re heading toward powerful personal assistants that are fully private.

[CG @cgtwts]

This guy just got Qwen 3.5 running on-device on the iPhone 17 Pro

Very soon, most serious AI users will run powerful local models on their own devices.

[Joel Jacob Stephen @joeljstephen]

Full Breakdown: https://t.co/n6qVDXUmyE

[Joel Jacob Stephen @joeljstephen]

A lot happened in AI this February.

Agents got their own environments to work in. Affordable models rewrote what you can justify building. And the wider world pushed back in ways nobody quite expected.

My February roundup ↓

[Joel Jacob Stephen @joeljstephen]

Models from MiniMax, https://t.co/MxN2lyS5gy, and Kimi are starting to feel like the mid-range smartphone tier. You get 90% of the flagship experience for way less money.

[MiniMax (official) @MiniMax_AI]

Introducing M2.5, an open-source frontier model designed for real-world productivity.

- SOTA performance at coding (SWE-Bench Verified 80.2%), search (BrowseComp 76.3%), agentic tool-calling (BFCL 76.8%) & office work.

- Optimized for efficient execution, 37% faster at complex tasks.

- At $1 per hour with 100 tps, infinite scaling of long-horizon agents now economically possible

MiniMax Agent: https://t.co/aIzrFYcfUz
API: https://t.co/fHRdSV7BwZ
CodingPlan: https://t.co/FDhZBBjQrX

[Joel Jacob Stephen @joeljstephen]

I built a small homelab tool that monitors my YouTube playlist, summarizes each new video with an LLM, and emails me the highlights every morning. Now I can skim through them in minutes and decide what’s worth watching fully. It’s saved me a ton of time.

[Joel Jacob Stephen @joeljstephen]

This looks like the cleanest, most thoughtful agent orchestration UI so far.

[OpenAI Developers @OpenAIDevs]

We’re excited to launch the Codex app, a command center for building with agents.

It gives you a focused space to manage multiple agents at once, run work in parallel, and collaborate with agents over long-running tasks.

https://t.co/ldE9k0uL5z