Amazon is holding a mandatory meeting about AI breaking its systems. The official framing is "part of normal business." The briefing note describes a trend of incidents with "high blast radius" caused by "Gen-AI assisted changes" for which "best practices and safeguards are not yet fully established." Translation to human language: we gave AI to engineers and things keep breaking? The response for now? Junior and mid-level engineers can no longer push AI-assisted code without a senior signing off. AWS spent 13 hours recovering after its own AI coding tool, asked to make some changes, decided instead to delete and recreate the environment (the software equivalent of fixing a leaky tap by knocking down the wall). Amazon called that an "extremely limited event" (the affected tool served customers in mainland China).

There’s no reason why heavy ops companies shouldn’t have personalized software No software? document you’re entire process Process not documented? you need to create operating procedures If you don’t have operating procedures, start documenting

Google Sheets should only be a V1 for business operations. Meaning you should move to V2 once ideas are validated

done. going to start releasing all the micro tools I use to speed up my commercial real estate tasks. introducing C2C company2contacts.com/

anyone using zoho for email and a crm?

I just tried Monologue and I'm experiencing audio hijacking. Moving back to Wispr Flow

its been 3 months i’m sick to my stomach💔💔

Claude Code removed the todo list items from a recent update They are working on a fix to revert but in the meantime running this hook based repo in another terminal is a good workaround!

playing around with medgemma for a few clients...i didn't realize it was an SLM which apparently where things are headed. "The future of AI may be many small, specialized models working together, not just a few colossal “super” brains." arxiv.org/abs/2506.02153

everyone’s celebrating gpt-5’s “unified system” and claude 4.1’s “drop-in replacement” while missing what’s actually happening. all four labs just converged on the exact same architecture. thinking models with routers. extended reasoning with tool use. hybrid architectures that decide when to think hard. gpt-5 has its router picking between fast and reasoning modes. claude 4.1 has hybrid reasoning with extended thinking. gemini 2.5 has “thinking built in” by default. grok 4 heavy runs multi-agent collaboration. they’re all the same model now. this isn’t innovation. it’s convergent evolution. when everyone independently arrives at the same solution, it means we’ve hit a local maximum. the paradigm is exhausted. here’s the part nobody’s saying out loud: we’re 3-6 months from every frontier model being functionally identical. same benchmarks. same capabilities. same price points. gpt-5 at $15/mo, claude at $15/mo, gemini will price match. the only differentiator left? distribution. openai has microsoft and chatgpt’s 700m users. anthropic has amazon and the enterprise. google has… google. and grok has x’s firehose and elon’s reality distortion field. we’re watching the last real model race end in a four-way tie. what comes next isn’t better models. it’s who can ship the most compelling products on top of commodity intelligence. the ai model wars are over. the platform wars just began. @sama @demishassabis @AnthropicAI @elonmusk

its the first of many pricing structures…pay to your data private help.kagi.com/kagi/why-kagi/…

I live in iterm2

AI subscriptions I currently pay for: - ChatGPT, $20/mo - Claude Code, $20/mo - Cursor, $20/mo Gotta take advantage of the VC-subsidized flat-rate subscriptions while you can

"however, writing is not only about reporting results; it also provides a tool to uncover new thoughts and ideas." part of my goal of posting on here. nature.com/articles/s44222-0…

"Compounding Engineering" the next iteration? the next skill is learning to control claude code. anyone can vibe code low quality code...intentional improvements like this is what will set codebases apart.