Join free with your email
Free. No spam, ever — we'll never share your email address and you can opt out at any time. Already a subscriber? Log in
Tracked through public AI activity and peer connections inside the directory.
We’ve released Letterweave, a small AI model that identifies what an email is for, from receipts and newsletters to invitations and personal correspondence. The model and training dataset are public on Hugging Face. Try it and let us know what you think. huggingface.co/Groktop…
If you like the idea of Jev but want a self-hosted option, here's a free & open source project I've started contributing to. This lets you host the Laya model, and currently presents a Jev-shaped API contract. I'm working with the maintainer on offering more native Laya-shaped…
AI Agent Skills v0.5.0 is out: - evidence-aware life coaching - cyberpunk - llama.cpp - Grafana - deeper Raleigh civic-data workflows Now backed by reproducible, paired, rubric-based evaluations. github.com/magnus919/ag...
Some AI models can detect injected “thoughts” without being trained to do it. Only sometimes, but how did that ability emerge? Via @quantamagazine.org @benbenbrubaker.bsky.social I found this to be an interesting read.
There’s a lot of BS in AI right now but there’s also some brilliant thinking out there that can fall under the radar. OpenRig thinking is now on my radar. The idea of an “everything factory” is making a lot of sense.
Just got to New York for my talk "Dark Factory Considered Harmful" at AI DevCon. I see this as a trend fraught with misinformation, which shows how hype distorts potentially useful ideas. In our new podcast we unpack how this fits in the bigger picture. www.thecodingagency.org…
The Ghost of Software Factories Yet to Come offers a warning, not a prophecy. If AI does more of the coding, do people get time and agency back, or just higher quotas?
The Ghost of Software Factories Present has agents writing code. But who decides what gets built, and who can stop the line when it's wrong?
Giving an AI agent access to payroll makes “guardrails” feel a lot less abstract. Who sets the limits, and who cleans up when it gets things wrong? That’s what I’m digging into in this week’s Sunday Signal.
AI makes you faster. Who gets stuck cleaning up? My conversation with @grimalkina.bsky.social pushed me to ask harder questions about fairness, invisible work, and whether the people governing AI have earned our trust.
Would you work for a company run by AI? Who would you appeal to when your AI boss gets something wrong? Delaware’s proposal for AI-run companies sent me digging into businesses already trying this, the humans working inside them, and the science fiction that saw it coming.
If AI gives a nurse more time with patients but doesn’t improve profit margins, has it failed? Healthcare leaders: where does the saved time go, and who gets to decide? My latest Groktopus newsletter: www.groktop.us/healthcare-h...
I love(d) Hermes Agent but TBH it's not been aging well for me. A lot of the scheduled jobs I've set up for it start falling apart when python updates happen. And it seems helpless to repair itself when it runs into this. Curious what open source AI agents are up and coming next gen.
Where, oh where, is the pro AI community on Blue Sky? How do I find them? Where are the people that are actually doing cool things with harnesses and skills and who don't think this is the end of the world? I tried finding them on X, and what I found was so toxic it leaves me ill.
I don't think that AI is going to rise up and kill us all, but I have noticed that newer frontier models are getting far more opinionated and very willing to strongly disagree with me on things that it's writing under my name.
At some point, I felt like digital cameras surpassed what I was getting out of film, and I made the switch. I kept all my old cameras, but now photoshop & AI have taken the fun out of digital photography. At some point, I might like to shoot film again, just not today.
I'm starting to catch other AI newsletters pinching off mine, like right down to the signature aesthetic.
The DJI Wireless mic is the hot setup for people who use Wispr Flow in their work every day. This is proving to be really useful for having voice-to-text sessions with my AI agent.
Just to say it again... don't trust ChatGPT or Claude with editing or refining prose. This is something Gemini is 1000x better at.
Eventually the hardcore anti AI folks are going to be pariahs. The more of them end up having good experiences with modern tools, the less dogmatic the more their voices will thin out.
Center = magnus919.com. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.
Add the Who’s Who of AI badge to your site or bio. It links back to this profile.
Markdown: [](https://aiweekly.co/whos-who/person/magnus919-com)