@ZryMiller I use gpt 6.1 sol to do code reviews and look for problems in my opus 5.5 code. It does amazing at that one specific thing, finding 3x the number of problems than an adversarial opus 5.5 review agent.
Day 3 of my M5 Max PROLED XDR @Omarchy bring up.
Touch ID, the camera and all three USB-C ports work.
Reading the fans and 335 sensors led to a power fix: idle 74 W to 20 W. And dropping.
Next: speakers, external displays and the ANE.
Details: https://t.co/ACZ2v6ZZu7
@jans So, can this do what Hermes or grok bot does, which is basically assign agents as specialists with their own context and memory, and have an orchestrator? Hermes is shit, and grok bot I can’t choose the model, but otherwise works really well.
@RileyRalmuto A real question, if you don’t mind, while we’re talking about agents and memory: I tried Hermes, even spend hours with Claude and codex fixing bugs, but can’t get it to do anything nearly as smoothly as grok bot, stops all the time, inter-agent communication problems & so slow —
@ErickAlvarezDev@JoshuaSWarren I rather would like to buy another Mac, thank you. It’s the best hardware. I don’t mind macOS for the most part either. It’s just not very good for the development level I do. So I’m quite happy about this project. Thanks for the suggestion though.
@badlogicgames If you haven’t yourself tried a rewrite or super complex projects w/ the new models (opus 5.5/fable 5.2 for writing & planning, & OpenAI gpt 6.1 sol as tester/debugger), I’d be wary of having a strong opinion against ai. In the hands of a talented engineer, it’s wildly powerful.