Getting Started with Grok Bot
01:00 What is @Bot?
05:26 Anatomy of a bot
08:49 3 ways to run
12:04 Multi-bot chains
16:06 Workflow 1: X followers to @Notion
20:29 Workflow 2: Personal software
23:08 Workflow 3: Coding with @cursor_ai
26:44 Workflow 4: Search
28:27 Recap
Ladies and Gentlemen…boys and girls of all ages…we have stable video output with the bare metal Omarchy install on the M4 Air, native GPU acceleration . Audio should be tackled today.
#OmarchyMac#OmarchyLinux
Apple locks fast local AI to macOS with Metal.
I picked the lock.
metal2vk translates Metal GPU code to Vulkan, unchanged.
598 of 610 real programs from @trymirai's uzu and @ashxhart's TensorFold so far.
Engines built for Macs now run on Linux.
https://t.co/Hl5k2EDxAT
One of the main lab goals this weekend is to tune power consumption and management on Apple Silicon.
It needs to be as good as MacOS.
I am hopeful of good news soon on this space.
@dhh@omarchy@OmarchyMac
I just built Word, Excel, and PowerPoint into Herald OS. All powered by Hermes Agent
And you can talk to all three.
"Make this document more professional." Hermes rewrites it.
"Calculate my monthly expenses." Hermes builds the formulas.
"Turn this document into a presentation." Hermes creates the slides.
Opens and saves .docx, .xlsx, and .pptx files.
No subscriptions. No switching between apps.
Free and open source.
https://t.co/PhImN5QgUT
Thanks to a recent @sudoingX post , I have put my Strata tests on hold until more RAM arrives.
Bucoid/Qwen3.8-27B-Heretic-Ara-16GB-VRAM-IQ4-XS-MTP-GGUF on a RTX 5060 ti. Solid 50+ tok/s with 64k context.
Grok Bot could release the exact feature you need, and you could carry on doing things the hard way for weeks.
Here’s a routine I’d set up so that doesn’t happen.
Once a week, have Grok Bot compare its latest updates with how you actually use it.
Your routines. Your connected apps. Your workarounds.
The things you keep complaining about.
Give it this prompt:
“Create a weekly routine for Monday at 8am.
Read the official Grok Bot release notes since your last successful check. On the first run, cover the previous seven days.
Compare those changes with my current setup and recent problems, using the context you can access. Flag anything you can’t verify.
Recommend up to three changes worth considering.
For each one, tell me:
1 - Which specific job or problem it helps with.
2 - How I handle that job today.
3 - What the update could improve or replace.
4 - The smallest test we could run to check whether it’s better.
Include the source for each recommendation.
If nothing is relevant, say so. Don’t fill the list.
Don’t change my setup without approval.”
The useful output would look like:
“You built this workaround because of a limitation. There’s now an update worth testing against it.”
That’s the connection worth automating.
A product update only becomes valuable when it improves something you actually do.