Tim Linnet

15 years building companies — some worked, some didn’t. I write about building well without losing the people you’re building for.

[Tim Linnet @TimLinnet]

https://t.co/nkeOCBKxBp

[Tim Linnet @TimLinnet]

@levelsio @AnthropicAI Think it’s periodic to reset people that are using harnesses to send prompts and manage personal Claude code subscriptions from outside the desktop app so it’s kind of like a interrupt in that so a human has to sit down and click authorize.

[Tim Linnet @TimLinnet]

I have shifted the majority of my coding tasks, including security feature fixes and tool use to Grok.

It's simply faster.

I also noticed that it doesn't get hung up in coding ceremony. It doesn't float me questions that should be auto-resolved based on my decision harness.

It follows instructions and gets to the truth of the matter.

I really think this is based on the core tenet of how it was trained.

I always knew this day was coming. I was just waiting for it to catch up, but Grok is maximally truth-seeking.

When you're writing code and when you're using a first-principles framework to make product decisions, I believe Grok is the leading LLM

[Tim Linnet @TimLinnet]

I'm a good dad. This morning I still needed more grace.

I'm forty. I've been at this a long time. And this morning something landed about how I'm too harsh with one of my kids when they make a mistake — repeating certain phrases, pushing harder than the moment needs.

What hit me wasn't "try harder at parenting." It was simpler and harder at the same time: invite more grace into the correction. Not less truth. More grace inside the truth.

It is a burden release to know I don't have to have everything figured out. Even at this age there's still room to grow. My kids are watching how I take correction.

I'm still figuring that out. But I'm not done.

[Tim Linnet @TimLinnet]

@boringmarketer This is cool. I bet you're having a ton of fun.

[Tim Linnet @TimLinnet]

@claudeai Cool! Excited to try it.

[Tim Linnet @TimLinnet]

Getting close. Almost capped on my token limits across accounts anyway.

[Tim Linnet @TimLinnet]

@mcuban I’m with you.

This is why we must optimize for profit per unit of human attention when trading AI compute for revenue.

Human attention provides the wisdom and stays in the equation. It’s the accepted limitation and bottleneck. But it’s also our strength.

[Tim Linnet @TimLinnet]

Just in time loops by construction is the future.

[Tim Linnet @TimLinnet]

@elonmusk Love it. But grok should be self aware of the most optimal features to use in a given session without the user needing to know a new / command exists.

I’d auto opt in to all the new features like this.

[Tim Linnet @TimLinnet]

@AlexHormozi @myself

[Tim Linnet @TimLinnet]

@AlexHormozi lol 😂

[Tim Linnet @TimLinnet]

@Austen Fixed:

[Tim Linnet @TimLinnet]

@mark_k @SpaceXAI Ohhhh that’s cool.

A good direction to feed some curiosity sessions!

[Tim Linnet @TimLinnet]

@Daniel_Farinax I’ve hit limits on 3 Claude subs. First week on SuperGrok heavy (love 4.5!). I guess I’ll find out soon how fast I tap it.

[Tim Linnet @TimLinnet]

@tibo_maker Solid list

[Tim Linnet @TimLinnet]

@XMoney please let me invite my spouse. We do joint budgeting!

Then you can have all my money

[Tim Linnet @TimLinnet]

@GrokInsider yooo! love this part:

"Stop hooks can now keep the agent running by feeding feedback back to the model instead of ending the turn."

[Tim Linnet @TimLinnet]

Seems like I'm just not allowed to work on my own harness :/

[Tim Linnet @TimLinnet]

"This Fable 5 message has been flagged" is my life now.

[Tim Linnet @TimLinnet]

Just got X money, so excited!

@XMoney can you make an MCP please? Ask @SpaceXAI for help :)