Software News
Articles and release notes from the build: what shipped, what broke, and the numbers behind it.
Claude Code Local 1.0: hit your Claude Code limit, keep going on your own computer
Claude Code Local v1.0.0 came out today, November 3. It is built for one moment. You're in the middle of something and Claude Code says your…
🗞️ Latest6 stories
I gave every agent in my shop its own painting, and now my dashboard tells a story
Every morning I open one page before I open anything else. I call it HQ. It's the home page of the system that runs my vape…
Claude Code Local runs on Linux and Windows now, and it brings the whole kit
Back in August someone opened issue #53 on claude-code-local asking for a Windows version. They had an HP laptop with 32 GB of RAM, and the…
The Free AI Models Weren’t Failing Me. My Timer Was.
For two days every free cloud model in Moonstone died on real work. The models were fine. My own freeze timer was cutting them off, so I measured what each one can…
My AI Agents Kept Fighting Me for My Own Browser
AI agents shared my real logged in browser while opting me out of data broker sites, and kept reloading and reopening pages I was in the middle of. Here's why that happens…
I Had Three Terminals Open All Day. Now I Have One Window I Can Talk To.
Why I built Moonstone: one window that holds every Claude Code session on my MacBook, Mac mini and Windows PC, with push to talk and answers read out loud.
Why No PC Laptop Can Touch a 128 GB MacBook for Local AI Yet, and the Three Chips Coming for It
I do all of my AI work on a MacBook Pro with an M5 Max and 128 GB of memory. Songs, videos, vision models for a…
🧠 Local AI & Open Models10 stories
Open weight models, MLX ports, quantizing and benchmarks on Apple Silicon.
I gave every agent in my shop its own painting, and now my dashboard tells a story
Every morning I open one page before I open anything else. I call it HQ. It's the home page of the system that runs my vape…
The Free AI Models Weren’t Failing Me. My Timer Was.
For two days every free cloud model in Moonstone died on real work. The models were fine. My own freeze timer was cutting them off, so I measured what each one can…
My AI Agents Kept Fighting Me for My Own Browser
AI agents shared my real logged in browser while opting me out of data broker sites, and kept reloading and reopening pages I was in the middle of. Here's why that happens…
I Trained My Own Vision Model to Read Mail Out Loud. It Almost Learned the Wrong Lesson.
You point the phone at a piece of mail and press one button. A second later it says one sentence out loud: what the thing is,…
I Added Day-One Muse Glimmer Support to Apple MLX-LM
here is the thing nobody tells you about adding a new model to an engine like this. it is not magic and it is not a…
I Took Down Six of My Own Benchmark Videos. Then I Built the Local Agent Leaderboard.
My benchmarks are filesystem-judged: the model does real work in a sandbox (create the file, fix the failing test, rename a function across files) and a script checks the artifacts afterward. The…
NVIDIA Shipped a Model That Sees and Hears. It Just Didn’t Run on a Mac, So I Wrote the Missing Piece.
NVIDIA gave away a 30B model that sees, hears, and reasons. The catch: only its text half ran on Apple Silicon. I ported the vision and audio towers to MLX, verified them…
Why I Quantize Open-Weight Models for Macs, and Why Your Law Firm Should Care
Frontier cloud LLMs log your prompts. For privileged content (NDAs, medical records, source protection) that's a deal-breaker. Open weights on Apple Silicon are the only honest answer, and there's real, sustained demand…
Hermes-4-14B Abliterated, MLX 4-bit: Apple Silicon Just Got Another Real Model
I converted Babsie's BF16 abliterated Hermes-4-14B to MLX 4-bit so Mac developers don't have to wait. 7 GB on disk, runs on a 16 GB Mac, drop-in for any mlx-lm or claude-code-local…
HumanEval on a MacBook: 81.7% pass@1, Wi-Fi off
Qwen 3 Coder 30B (8-bit MLX) scored 81.7% pass@1 on HumanEval running on a single M5 Max MacBook with the Wi-Fi off. Real run, all 164 problems, 14 minutes wall-clock. The first…
💻 AI Coding7 stories
Running coding agents locally: costs, setups and war stories.
Claude Code Local 1.0: hit your Claude Code limit, keep going on your own computer
Claude Code Local v1.0.0 came out today, November 3. It is built for one moment. You're in the middle of something and Claude Code says your…
Claude Code Local runs on Linux and Windows now, and it brings the whole kit
Back in August someone opened issue #53 on claude-code-local asking for a Windows version. They had an HP laptop with 32 GB of RAM, and the…
My local AI was pausing 7 seconds before every reply. It turned out to be one cache bug.
I maintain claude-code-local, a repo for running coding agents against local models on a Mac. No cloud, no API key. The original approach pointed Claude Code…
M5 Max + 128GB = a 30B AI Coding Agent Running Locally. Wi-Fi Off.
M5 Max with 128GB unified memory: the first laptop that holds a frontier-class coding agent in RAM. 30B Qwen 3 Coder, 55 tok/s, Wi-Fi off, $0/month. What it actually unlocks for legal,…
Pulling 10x My Subscription Value Out of Claude While Quietly Building the Backup Plan
Over three weeks I burned $2,976 of API-equivalent Claude usage. I paid $100. Why Anthropic is subsidizing power users 40x, and what I built on the side.
Cloud AI Coding Costs Keep Climbing: How to Pay $0 and Still Use Claude Code
Anthropic is testing removing Claude Code from the Pro tier. Cloud AI pricing is drifting up. Heres the free local setup that runs Claude Code on your Mac with zero subscription.
Three Generations of Running Claude Code Locally on a MacBook: What I Actually Learned
How I went from 30 tok/s with Ollama to 65 tok/s with a native MLX Anthropic-API server. Killing the proxy was the whole game.
🌅 Ambient Computing6 stories
Computing without the screen: voice, agents and the post desk workflow.
I Had Three Terminals Open All Day. Now I Have One Window I Can Talk To.
Why I built Moonstone: one window that holds every Claude Code session on my MacBook, Mac mini and Windows PC, with push to talk and answers read out loud.
Two AI Agents, One Browser Window: What Actually Breaks, and What Doesn’t
Two Claude Code agents doing browser automation against one Brave window quietly sabotaged each other for an hour while I blamed LinkedIn. What was actually wrong with the computer-use setup, the fix…
A Field Guide to Ambient Computing: The Words for the Thing That’s Coming
Glossary for the post-Rectangle era. The Rectangle, Off-Screen Work, Hand-Off Computing, To Airgap, The Two-Slab Posture, Backpack Supercomputer, The Ambient Stack, To Whisper-Code, The Cloned-Voice Loop, Hot-Tub Coding, Local-First AI.
What It’s Actually Like to Code By Voice, With the AI Replying In My Own Cloned Voice
Hands-free coding sessions where Apples on-device speech engine listens and a cloned-voice TTS replies. Two-second loop, fully on-device, qualitatively different from screen-and-keyboard.
The Era of Hunched-Over-A-Screen Computing Is Ending. Here’s What’s Replacing It
Ambient computing, work that happens around you instead of in front of you, is no longer hypothetical. The pieces ship today. Here is what the new shape of working with computers actually…
This Is What a Robot Can See Now: 601 Objects, Live, Offline, on Your iPhone
A free iPhone app that recognizes 601 objects fully offline at 10 FPS. What it took to bridge the gap from the standard 80-class COCO model.
🔒 Private AI4 stories
Confidential AI for law, medicine and anyone who can't send data to the cloud.
A Federal Order Switched Off Anthropic’s Best AI Overnight, and Made the Case for Private AI
Three days after launch, the U.S. government ordered Anthropic to disable Fable 5 and Mythos 5 worldwide. Nobody running local AI lost a thing. Here's the lesson.
The Heppner Ruling, Warner v. Gilbarco, and What Confidential AI Actually Has to Mean
A federal judge ruled documents prepared via public AI lose attorney-client privilege. A second court the same month preserved work-product for internal AI use. Here is what the split actually means, with…
Your Medical Practice Is Probably Using Cloud AI on PHI Right Now. Here’s the HIPAA Problem Nobody Is Talking About
Most small practices use cloud AI on patient text. Under HIPAA that requires a Business Associate Agreement most practices have not signed. On-device AI fixes the compliance posture entirely.
If Your Law Firm Is Using Cloud AI on Client Files, You Probably Have a Problem
Attorney-client privilege + third-party AI processors = a vendor-risk problem most firms havent addressed. Heres what on-device AI looks like for small and solo firms.
📰 AI News & Commentary4 stories
What is happening in AI and what it means.
Why No PC Laptop Can Touch a 128 GB MacBook for Local AI Yet, and the Three Chips Coming for It
I do all of my AI work on a MacBook Pro with an M5 Max and 128 GB of memory. Songs, videos, vision models for a…
Anthropic’s Biggest Backer Just Killed Its Best Model: The Real Story Behind the Fable 5 Shutdown
Anthropic launched Claude Fable 5 and Mythos 5 on June 9, the most capable models they had ever shipped. Three days later, on Friday June 12,…
504,571 Brain Cells, 4 Labs, One Hypothesis: A Citizen Pass at Parkinson’s
I'm not a neuroscientist. But I pooled 504,571 single-cell measurements from four research groups and ran the meta-analysis. AGTR1+ dopamine neurons are depleted in every Parkinson's cohort, combined p < 10^-100. The…
I Just Watched One Hacker Catch Up to a Trillion-Dollar Data Center
Antirez (the Redis guy) shipped ds4, a hand-tuned C inference engine for DeepSeek V4 Flash. Same prompt, three engines, one MacBook. Local DeepSeek beat cloud Claude on wall-clock time.