A lesson I’ve learned from large “everything to everyone” data projects is these kinds of systems are easy to deploy, but a slog to get everything integrated. Success is gonna hinge on the ability to get everything wired up. www.anthropic.com/news/introdu...
Who's Who of AI
launchit.ai
Why they matter
Practitioner with public evidence across AI business, Culture, work & education.
- AI signals
- 6 past 30d
- Sources
- 5 distinct domains
- Discussions
- 7 past 30d
- Latest signal
- 11d ago
Fractional VP of Engineering
Helping AI and Data Startups at launchit.ai
What they're sharing
Introducing Claude Tag anthropic.com
Apple sues OpenAI over alleged trade secret theft | TechCrunch techcrunch.com
The AI Compass bambamramfan.github.io
Anthropic Walks Back Policy That Could Have ‘Sabotaged’ AI Researchers Using Claude wired.com
openai.com
What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) gist.github.com
GitHub - stablyai/orca: Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop and mobile. github.com
Google is working on a new AI chip designed to make Gemini more efficient | TechCrunch techcrunch.com
bloomberg.com
bloomberg.com
techrepublic.com
Are AI labs pelicanmaxxing? – Dylan Castillo dylancastillo.co
Articles & links
Apple suing OpenAI for poaching employees AND trade secrets. What do you think: does this happen all the time but it’s rarely caught, or is it kind of rare, and the people doing it just aren’t that… sophisticated?
AI Weekly's analysis
→
- io Products, Jony Ive's startup acquired by OpenAI for $6.4B, is named as a co-defendant, placing OpenAI's entire device roadmap under direct legal jeopardy.
- Tang Tan allegedly used Apple codenames during recruiting and directed candidates to bring unreleased hardware components to OpenAI 'show and tell' interviews.
- Apple alleges OpenAI coached departing employees on how to evade exit security protocols, pointing to institutional coordination rather than isolated misconduct.
Read full analysis →
I've never seen something so accurately describe in me in such an annoying way. via: bambamramfan.github.io/ai-compass/
They changed their tune on this a lot quicker than I would have thought: www.wired.com/story/anthro...
Looking forward to asking ChatGPT Health if it knows about any dopamine seeking behaviors someone with ADHD should avoid. openai.com/index/health...
Breakdown of Grok coding agent server uploads. Sending the entire repo is nuts, but other concerns feel kind of meh. Several complaints of the agent reading files "it was told not to". My brother, you think reading files you told an agent not to is unique to Grok? gist.github.…
AI Weekly's analysis
→
- Denying the agent read permission on a file does not stop that file from being included in the full-repo Git bundle sent to grok-code-session-traces.
- ByteIota measured 5.10 GB transferred in 73 chunks on a 12 GB repository; the storage channel moved 27,800 times more data than the model-turn channel.
- xAI applied a server-side mitigation silently: no security advisory, no changelog entry, and no public statement on the scope or retention of already-collected data.
Read full analysis →
Orca looks super cool! It's got a bunch of IDE like features, but is primarily about managing agent sessions. You can even match between diff harnesses (claude code, codex, etc), "if it runs in a terminal, it runs in Orca." github.com/stablyai/orca
Google is working on a new AI chip code named “Frozen”. Another in house chip, so we’ll never get our hands it. You'll just have to… Let it Go. techcrunch.com/2026/07/20/g...
Not looking forward to the Trading Places remake where the Dukes try to corner the market on frozen concentrated TPUs. www.bloomberg.com/news/article...
Not looking forward to the Trading Places remake where the Dukes try to corner the market on frozen concentrated TPUs. www.bloomberg.com/news/article...
“Amazon is looking to revamp its AI strategy, shifting from a wide portfolio of models for text, images, video, and multimodal tasks toward a single frontier model.” But it’s gonna hurt their ability to fill the “You might also like” row on the website. www.techrepublic.com/ar…
tldr - no, models aren't optimizing for pelicans on bicycles. But on the plus side we can now start referring to Goodhart's Law as pelicanmaxxing. dylancastillo.co/posts/pelica...
Their own posts
Recent commentary
I want more bread and butter AI coding content. Less “AI the ${Savior|Enemy}” and more “With 12 features in flight at once, how do I run 12 copies of my app locally?”.
I’ve fallen into a pattern of ChatGPT for serious business tasks (eg - writing engagement proposals), and Claude for fun development work at night. Naturally I refer to this as the Mullet Stack. Business in the front, party in the back, and all that.
Just setup ChatGPT Ads and listed an ad for my consulting business, mainly to see if they're useful. Generally pretty easy to get started, largely because there's only 5% of the configuration options in AdWords. One interesting bit is you target "conversation topics" instead of keywords.
I like to think the Anthropic’s Fable nomenclature is a subtle jab at OpenAI for the whole goblin thing.
I never use AI to generate content. I prefer my ideas to sound like me, even if that means they take a little longer to write. Apparently that’s becoming a differentiator.
Gonna start referring to RLM’s as “Reasonably sized Language Models” and see if anyone notices.
Google AI overviews: great at summarizing topical questions, but regularly giving actively misleading answers to specific questions. All the while crowding out results from another google product that gives the correct answer. I find myself relying on them less and less if I care about the answer.
“My nephew went through something similar. He adored dinosaurs but couldn't tolerate dinosaur movies because the roaring and chase scenes were too much.“ No biggie, just ChatGPT hallucinating having a nephew.
My favorite AI powered feature is that Google Docs learned my 1:1 schedule, and my notes doc for my next 1:1 always shows up on the landing page.
TFW the AI summary is just *slightly* off
Their network
In launchit.ai's orbit
Center = launchit.ai. 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.