The Computomatix Times

github.comGitHub - mizorewww/laya-mlx: Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API.Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API. - mizorewww/laya-mlx

Open-Source Laya-MLX Runs Fast Decision Models on Apple Silicon

▶

A developer introduces laya-mlx, an open-source MLX port of the Laya typed decision model, claiming 7-14 ms decisions on an M3 Max and a demo playing Snake at 60 decisions per second. A GitHub repository is linked.

雨夹雪❄️ @mizorewww ·Sep 19, 2026 ·♥ liked

Original post · 1 min read

介绍比Jev快50倍，在你设备上跑的laya-mlx!

只在你的设备上占用最高1G内存

Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统

我将其移植到MLX,并且做了一些性能优化!

视频中就是这个模型在我的本地M3Max上玩贪吃蛇

这个模型能够以每秒决策60次的速度玩贪吃蛇！

github.com/mizorewww/laya-mlx

♥ 14.1K · ⟲ 1.2K · 👁 4.3MView on X ↗

LuxAlgo Lets Pine Script Indicators Run Outside TradingView

LuxAlgo announces that Pine Script indicators can now run outside TradingView without rewriting, including its Smart Money Concepts indicator, usable in other apps, charts or agents. A commenter predicts TradingView's decline.

The Milk Man @MrMilkTrading ·Sep 29, 2026 ·♥ liked

Original post · 1 min read

TradingView is cooked, only a matter of time

The BlockBuster of the trading world

LuxAlgo @LuxAlgo
For the first time in history, Pine Script® indicators can run outside of TradingView.

No rewriting. No Python port. Paste the same code you already have into your own app, your own charts, or your agents, and it runs 1:1.

Left: Smart Money Concepts, the #1 indicator on TradingView. Right: same code, our charts.

Thread on how to use it yourself: 🧵

♥ 1.3K · ⟲ 66 · 👁 191.4KView on X ↗

Fastbrowse Launches as Open-Source, Low-Cost Browser Agent

Furqan Rydhan introduces fastbrowse, an experimental open-source browser agent where an LLM plans actions and each claim cites an exact quote from the page. He says it is significantly cheaper and faster for agents to browse the web.

Furqan Rydhan @FurqanR ·Sep 20, 2026 ·♥ liked

Original post · 1 min read

Introducing fastbrowse.

An open-source fast browser agent.

Jev picks each action directly from the page, an LLM plans and every claim cites an exact quote.

It's significantly cheaper and faster for agents to browse the web now.

Early and experimental, but very promising.

fastbrowse.ai

♥ 578 · ⟲ 37 · 👁 71.9KView on X ↗

macOS 27 Adds Command-Line Access to Apple Foundation Models

Taha Tesser reports that macOS 27 includes a new fm command-line tool that lets users run Apple's on-device Foundation Models from the terminal.

Taha Tesser @TahaTesser ·Sep 20, 2026 ·♥ liked

Original post · 1 min read

TIL macOS 27 now comes with new fm cli 🤯

We can now use Apple's Foundation Models (on device) from CLI!

♥ 1.2K · ⟲ 72 · 👁 282.2KView on X ↗

Patrick OShaughnessy Highlights DHH Talk on Code Writing Shift

Patrick OShaughnessy praises a talk in which DHH argues that writing code by hand is no longer economically viable for most programmers, drawing a parallel to how technology reduced photo-taking and predicting the same progression for code.

Patrick OShaughnessy @patrick_oshag ·Sep 24, 2026 ·♥ liked

Original post · 1 min read

This talk is so, so good

Favorite idea—This photo shows what tech did to number of pictures taken.

Now same progression happening to code.

Then it’ll happen to…

DHH @dhh
It's pencils down, people. Writing code by hand is no longer an economically viable skill for most programmers at most companies. But the future of making software has never been brighter. Don't you dare black pill this beautiful moment! youtube.com/watch?si=6FsfJXS22mk52gyu&v=vDjW_d…

♥ 538 · ⟲ 41 · 👁 134.6KView on X ↗

DHH Declares Hand Coding Economically Obsolete for Most Programmers

David Heinemeier Hansson argues that writing code by hand is no longer an economically viable skill for most programmers at most companies, a view Karthik Hariharan says most engineers saw coming and notes raises questions about hiring and training.

Karthik Hariharan @hkarthik ·Sep 25, 2026 ·♥ liked

Original post · 1 min read

Most of us saw this coming over the last year, but @dhh weaves the narrative well. Hand coding is likely done for most software engineers.

We still have to figure out how to train and hire new software engineers though so maybe hand coding will live on in leetcoding for a while.

DHH @dhh
It's pencils down, people. Writing code by hand is no longer an economically viable skill for most programmers at most companies. But the future of making software has never been brighter. Don't you dare black pill this beautiful moment! youtube.com/watch?si=6FsfJXS22mk52gyu&v=vDjW_d…

♥ 48 · ⟲ 1 · 👁 7.7KView on X ↗

DHH Says Agents Make Trying Ideas Cheaper Than Planning

DHH argues that the opportunity cost of endless planning has risen because AI agents let builders try far more things, and says people learn faster by letting agents experiment.

DHH @dhh ·Sep 26, 2026 ·♥ liked

Original post · 1 min read

This has never been more true. The opportunity cost of endless planning and contemplation just went way up. Agents allow you to just try vastly more stuff, so let them, and you'll learn way more, way faster.

♥ 5.3K · ⟲ 468 · 👁 255.2KView on X ↗

Garry Tan Recommends Aside Browser for Agents Facing Anti-Bot Barriers

Garry Tan recommends running the AsideAI browser with MCP on a spare, always-on computer so agents can use real credentials inside a real Chromium browser. He calls it a gamechanger for getting past antibot friction.

Garry Tan @garrytan ·Sep 27, 2026 ·♥ liked

Original post · 1 min read

If you like Muse and Grok Bot but still run into crazy antibot annoyances try @AsideAI browser with MCP with your agent on a spare laptop or computer you keep plugged in somewhere.

It lets your agents your real credentials as yourself from a real Chromium. Gamechanger.

♥ 1.2K · ⟲ 59 · 👁 151.5KView on X ↗

claude.comAgentic coding is straining CI. Here’s how we scaled test impact analysis at Anthropic | Claude by AnthropicOur CI job volume increased 25x over 6 months. We patched our test selection service three times before finding a sustainable solution.

Sabba Keynejad Demonstrates AI Video Editing With OpenEdit

▶

Sabba Keynejad posts a video showing Opus 5.5 and the OpenEdit tool editing a fake tennis interview, presenting it as a new way to edit video.

Sabba Keynejad @sab8a ·Sep 30, 2026 ·♥ liked

Original post · 1 min read

This unlocks a whole new way of editing video

Gave Opus 5.5 a fake tennis interview... 😂

OpenEdit did the rest 👇

What do you think?

♥ 20 · ⟲ 0 · 👁 1.8KView on X ↗

Agents & Dev Tools8/10

Anthropic Says Claude Writes 80% of Its Code, Strains CI

Addy Osmani reports Claude now writes 80% of Anthropic's code and engineers ship 8x more per quarter. The side effects include 10x more tests and a 25x rise in CI jobs over six months, which Anthropic addressed with a scaled test impact analysis.

Addy Osmani @addyosmani ·Sep 14, 2026 ·🔖 bookmarked

Original post · 1 min read

At Anthropic, Claude now writes 80% of our code. Engineers ship 8x more code per quarter.

Side effect: Tests grew 10x. CI jobs up 25x in 6 months. Here's what helped us scale:

claude.com/blog/agentic-coding-is-straining-ci…

♥ 5.0K · ⟲ 367 · 👁 789.3KView on X ↗

Developer Lauren Shares Method for Shipping 2,500 PRs Monthly

▶

Developer lauren (@poteto) posts a free video walkthrough of how she shipped 2,500 pull requests to production last month. The talk was originally planned for Cursor Compile in London and is viewable on X at 2x speed.

lauren @poteto ·Sep 21, 2026 ·♥ liked

Original post · 1 min read

here's how i shipped 2,500 PRs last month to production

this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly

♥ 16.0K · ⟲ 1.4K · 👁 3.5MView on X ↗

DHH Says Models Can Also Handle Software Architecture

David Heinemeier Hansson responds to developers seeking to preserve architecture as a human domain, arguing that AI models are also very good at architectural work.

DHH @dhh ·Sep 25, 2026 ·♥ liked

Original post · 1 min read

I understand the appeal in trying to find the first plausible fortress in our retreat from writing code, but if you think it's "architecture", I have bad news for you. The models are also very good at that.

♥ 6.5K · ⟲ 280 · 👁 614.7KView on X ↗

Ahmed Khaleel Releases Native macOS Disk Scanner Built in Rust and Swift

▶

Ahmed Khaleel announces a native macOS disk usage app built with Rust and Swift that is 4.7 MB, scans 278,000 files in 0.7 seconds, and scans an entire 4-million-file Mac in about 13 seconds. He claims it is 1.4 times faster than disktree while using half the memory.

Ahmed Khaleel @ahmedkhaleel04 ·Sep 26, 2026 ·♥ liked

Original post · 1 min read

made a native macOS one (Rust + Swift)

→ 4.7 MB app
→ 278k files in 0.7 s
→ whole Mac (4M files) in ~13 s
→ 1.4× faster than disktree, half the memory

tobi lutke @tobi
Was wondering where my disk space went.
Therefore this exists now. You can simply wish software into existence.

♥ 906 · ⟲ 39 · 👁 135.7KView on X ↗

Hiten Shah Highlights Muse Permission Controls for Network Protocols

Hiten Shah shows a permission screen from the consumer AI agent Muse that exposes controls for SSH, SMTP, DNS, TCP and other protocols. He notes users can block protocols or require approval per connection and is hosting an AI Permissions 101 session.

Hiten Shah @hnshah ·Sep 23, 2026 ·♥ liked

Original post · 1 min read

Muse is a consumer AI agent.

This is one of its permission screens.

It exposes controls for outbound SSH, SMTP, IMAP/POP3, database connections, FTP, DNS, TCP and UDP.

You can block a protocol entirely or let Muse ask before each connection.

Friday at 10 AM PT I’m doing AI Permissions 101.

hiten.com/ai-permissions-101

♥ 137 · ⟲ 9 · 👁 18.1KView on X ↗

Kevin Rose Praises DHH's Candid Remarks on Writing Code With AI

Kevin Rose endorses David Heinemeier Hansson's comments on hand-written code versus AI-assisted coding, arguing that making past work obsolete is the purpose of building. He quotes Jay Ledev calling DHH a leading software artisan.

Kevin Rose @kevinrose ·Sep 26, 2026 ·♥ liked

Original post · 1 min read

Well said @dhh. I've 'retired' many times... from setting IRQ jumpers, defragging drives, and tuning config.sys for a few more kb of ram. making yesterday's work obsolete is the whole point of building. grateful to have watched it happen.

Jay Le @jayledev
DHH is probably one of the top software artisans of the past decade. Hearing him speak with this much conviction and transparency about one of the most polarizing topics in software right now, writing code by hand vs. with AI, is something pretty much every software engineer should hear a few times.

Especially those still on the wrong side of history.

♥ 492 · ⟲ 25 · 👁 76.7KView on X ↗

Shopify CEO Tobi Lutke Shares Tool for Creating Software From Wishes

Shopify CEO Tobi Lutke posts a photo with a playful note that he was looking for lost disk space, then announces a tool that lets users simply wish software into existence.

tobi lutke @tobi ·Sep 24, 2026 ·♥ liked · 🔖 bookmarked

Original post · 1 min read

Was wondering where my disk space went.
Therefore this exists now. You can simply wish software into existence.

♥ 7.8K · ⟲ 260 · 👁 1.3MView on X ↗

Tobi Lütke Argues Code Should Be Judged on Merit, Not Origin

Tobi Lütke responds to a report that KDE's draft AI policy discourages disclosing LLM use, saying code should be accepted on merit with a human accountable for it. He says good code is good and slop is slop regardless of how it was made.

tobi lutke @tobi ·Sep 20, 2026 ·♥ liked

Original post · 1 min read

This is the way. Accept code on merit and ensure that a person takes accountability for it. Doesn’t matter if it was typed, chiseled, generated, or bit-flipped via magnetized needle on a chip.

If it’s good, it’s good. If it’s slop, it’s slop.

The Lunduke Journal @LundukeJournal
KDE is working on an official AI / LLM policy, and it reads like the rules of Fight Club.

In short, KDE’s AI policy:

1) Encourages using AI, as long as a human is kept “in the loop”.

2) But you can’t tell anyone that you used AI.

“Don’t disclose LLM usage”.

“Don’t add ‘Assisted-by: [some LLM]” (as is done in the Linux kernel).

“Nobody in KDE should know if you use an LLM”.

In other words: “Welcome to developing KDE with AI. The first rule of developing KDE with AI is: you do not talk about developing KDE with AI.”

invent.kde.org/plasma/plasma-workspace/-/work_…

♥ 4.0K · ⟲ 228 · 👁 233.7KView on X ↗

Dex Horthy Argues Jev Fits Pipeline Design From 12-Factor Agents

Dex Horthy argues that Jev is a strong reason to revisit his 12-Factor Agents guidance, since tool calling can be split into classification and action. He recommends designing AI systems as pipelines that mix classification, structured data, deterministic code and small agent loops, citing a linked guide.

dex @dexhorthy ·Sep 17, 2026 ·♥ liked

Original post · 1 min read

jev is the best excuse you could possibly have to go re-read 12 factor agents. Tool calling itself can be decomposed into classify+action,

if you learn to design ai programs as pipelines that switch breathlessly between classification, structuring data, deterministic code, AND small agent-shaped append-chat loops, then jev is a WONDERFUL building block

hlyr.dev/12fa

Dillon Mulroy @dillon_mulroy
i think jev is resonating with devs so well b/c it unlocks so many opportunities for composing ai into systems and products rather than ai _becoming_ the product/system

really does feel like it was a missing primitive

♥ 2.7K · ⟲ 137 · 👁 233.6KView on X ↗

Full Context Highlights Open Source Jev Tool for Fast Browser Agents

Full Context introduces Jev, an open source tool from Browser Use that reviews a page and picks the next action, such as clicking, typing, scrolling or retrying. The author suggests splitting work between a large model for planning and a small fast model for choosing clicks, and notes the claims are still being verified.

Full Context @fullctx ·Sep 17, 2026 ·♥ liked

Original post · 1 min read

Just found Jev.

Jev looks at the page, sees the real buttons, and just picks: click this one, type here, scroll, wait, or done.

- Click this
- Scroll down
- Type this
- Buy / sell
- Retry or stop
- Send the job to another agent

That’s it.

This is useful for:
• Browser agents that need to be fast
• Auto QA
• Trading loops
• Agent routing
• Games / robots
• Anything where the AI has to choose an action over and over.

The real trick is splitting thinking from “what do I do next”. Big model does the hard thinking. And the small fast model just picks the next click.

Still looking into it, so forgive me for any false claims.

Open source here:
github.com/browser-use/jev-ultrafast

♥ 1.9K · ⟲ 160 · 👁 134.0KView on X ↗

danny.mdmotion-ad — danny.mdProduces short motion-graphics video ads (e.g.

Danny Postma Releases Skill for Generating 15-Second Motion Ads

Danny Postma announces a reusable skill that turns a single prompt into a 15-second motion-graphics ad for a product, linking to a skill page on his site.

Danny Postma @dannypostma ·Sep 26, 2026 ·♥ liked

Original post · 1 min read

turned this into a skill anyone can use

one prompt → 15s motion ad for your product

danny.md/skills/motion-ad

Danny Postma @dannypostma
joining the opus 5.5 hype, what a time to be alive

♥ 578 · ⟲ 19 · 👁 85.9KView on X ↗

Google Gemma Team Releases DiffusionGemma-Jev Endpoint on Cloud Run

Linus Ekenstam notes that frontier labs are rapidly releasing JEV forks, citing Google's Gemma team's djev. A linked Google post and GitHub repo describe deploying a Jev API-compatible endpoint on Cloud Run with one command.

Linus ✦ Ekenstam @LinusEkenstam ·Sep 23, 2026 ·♥ liked

Original post · 1 min read

I love seeing how fast the frontier labs are jumping in on creating forks.

Just today we’ve seen omni-jev and now djev from Google Gemma team.

It will be clear in a few weeks, just how powerful JEV is going to be inside harnesses and applications.

Google Gemma @googlegemma
Deploying DiffusionGemma-Jev (djev) just got a lot easier. You can now spin up a Jev API-compatible endpoint on Google Cloud Run using a single command.

Performance is solid: ~35-60 ms for single step latency and batch@32 is ~100-123 requests/sec.

It's a straightforward way to experiment without needing your own GPU. Runs at roughly $3/hr and drops to $0 when idle.

Get the code and instructions here: github.com/taeold/djev-run

♥ 67 · ⟲ 7 · 👁 18.2KView on X ↗

Free Open-Source Skill Adds Six Styles for App Screenshots

▶

Parth Jadhav announces six new styles for an open-source agent skill that generates app store screenshots, which users can select or match to their app's existing look.

Parth Jadhav @ParthJadhav8 ·Sep 26, 2026 ·♥ liked

Original post · 1 min read

The best skill to generate Screenshots for your apps is getting even better 🔥

We've added 6 new styles which users can pick from, and trust me - They're beautiful !

And ofc, you can just tell the agent to "use the style of my app"

+ ITS FREE & OPEN SOURCE 🧵

♥ 533 · ⟲ 33 · 👁 83.8KView on X ↗