Just a dude who really loves Grok Bot
Joined November 2023
- Tweets239
- Following402
- Followers330
- Likes385
Anyone else’s Grok Bot misspell all the time? or can’t even understand it’s output due to weird word choice?
So i’m loving Grok Bot, but can we get a small update where we can see our usage on the mobile app? @elonmusk
Ran the same build twice.
grok bot ate 20% of my usage bar.
opus 5 on ultra code mode did it in 4%. and it finished.
people keep saying grok is the cheap one. cheap per token is not the same thing as cheap per job.
if the agent burns a fifth of your month to get to the same page, the sticker price does not matter. @elonmusk
Claude felt revolutionary from January 2025 through roughly mid-2026.
Now I’m watching more people cancel their subscriptions entirely.
The model didn’t suddenly get dumb. The product around it did. Usage limits are terrible, and the gap between what you pay for and what you can actually use keeps getting harder to justify.
People don’t leave because the answers got worse.
They leave when the thing you paid for starts feeling like a ration.
I didn’t drop Claude because something else posted a higher benchmark.
I dropped it because I kept hitting a wall in the middle of real work.
You can forgive a model that’s imperfect. You cannot forgive a subscription that treats your afternoon like a meter.
That’s the whole story. The product got in the way of the intelligence.
Loyalty to a chatbot lasts until the friction outruns the output.
Claude had a long run. From early 2025 into this year it felt like the default.
Then the limits started deciding the day for you. You pay for access and get a countdown.
Once that happens, “better” is whichever one lets you finish the job.
The model is not the product.
The product is whether you can stay on a task without getting cut off, reset, or told to come back tomorrow.
That’s why people are leaving. Not a vibe shift. Not a leaderboard.
The bill stopped matching the work.
switched off the subscription I used to defend.
not because I found a smarter chatbot. because I got tired of planning my day around a usage cap.
if you’re still paying frontier prices for a rationed window, try grok 4.6 in cursor this week. it stays on the task. that’s the part that actually changed.
Most AI still waits for your next prompt.
Grok Bot doesn’t.
I closed the laptop last night with a half-finished research pile. This morning the agents had already sorted it, drafted the follow-ups, and left me a list of what still needed a human.
That’s the switch. Not a smarter chat. A worker that keeps going.
2.
People keep asking if Grok Bot is “another chatbot.”
It isn’t.
It has its own computer. It runs routines while you’re gone. You can split work across agents the way you’d split a real team.
If you’re still pasting the same context into a new chat every morning, that’s the part that dies first.
3.
I stopped treating agents like one long conversation.
I treat them like hires.
One watches the inbox. One does research. One writes. One just runs the loops I used to do by hand on Monday.
Grok Bot is the first setup where that didn’t feel like a demo.
If you already have skills and project files sitting in Claude Code or Codex, don’t rebuild from zero. Point an agent at the folder and tell it to turn that into roles.
4.
The underrated Grok Bot move is not asking it a better question.
It’s leaving it a job.
Write the outcome, the files it needs, and what “done” looks like. Then walk away.
Anyone who used it this week already knows. The ones still arguing about model scoreboards haven’t tried that yet.
5.
Quick prompt I keep using when I spin up a new Grok Bot:
“Read my notes and past work in this folder. Write a one-page brief of how I actually operate: what I care about, how I write, what I never want done for me, and which tasks should be standing jobs instead of one-off chats. Then propose 3 agents and the first routine each should run.”
Paste that once. You’ll get a team instead of another empty chat.
6.
I don’t need another model that writes a nicer paragraph.
I needed something that still works after I stop typing.
Grok Bot is the first one that felt like that. Own machine. Memory. Agents that don’t forget the project when the tab closes.
If your current setup dies the second you close the window, that’s the test.
Grok Bot is number one and people are still talking about it like a chatbot. It isn’t. You hand it a job. It has its own computer. It keeps working after you close the laptop. I switched because I got tired of models that wait for the next prompt. Grok 4.6 actually stays on the task.
Grok Bot Tips retweeted
switched my cursor default to grok 4.6. it actually stays on a long task instead of getting lost halfway through. that’s the whole reason I moved.
grok 4.6 is the first one that felt like I could leave it on a messy idea and come back to a working first pass. if you’re still paying frontier prices for the same work, try it in cursor this week. they doubled the included usage.
switched my cursor default to grok 4.6. it actually stays on a long task instead of getting lost halfway through. that’s the whole reason I moved.
Guys, Grok 4.6 is insane. It’s super cheap and the guardrails that Claude and Codex have are night and day difference. Allows you to be more creative. Get it. 4.7 will change the game as we know it