tarun

@bridge__harris might be in the wrong line of work with the GEO this strong...

I am Tarun now

@bartslodyczka @bcherny This!! Huge for security as well.

@bcherny This is very cool! Side question: will you guys introduce hooks for Cowork? I'd love an automated way to run a 'session close' skill at the end of each day. Thanks legend!

We've been working on this for a while. Can't wait to hear what you think

@trq212 claude scheduled tasks should have the ability to set which timezone when running. Had a task set to 9 AM CT, but didn't trigger in time because I was traveling in pacific time

@0xNairolf doing this with pressw

starting to see ai-as-a-service as an insane business - help people/companies find the best workflows to optimize - set up personalized ai agents for them - profit feels like the new consulting

ran clawdbot for about 5 minutes before it ran rm -rf on my downloads folder

@paulswaney3 INTEGRATION

@sweatystartup "The secret of change is to focus all of your energy not fighting the old, but on building the new" - Socrates

I've caught several of my employees running all of their communication through AI. We're on the verge of banning it in all of my companies. It isn't secure. It is a waste of time. And it is visible to clients, partners and team members from a mile away. Stop doing this!

How tf is it possible that my howie bot produced a typo when spelling my name?? @howie_ai @awwstn I checked the entire platform and my name is never misspelled. I checked my inbox and have ~10 emails where my name is misspelled. This is in contrast to thousands of emails with it correct. My only two guesses are that either: 1. The system prompt tells it to make mistakes on occasion so that it looks more human 2. This is not an AI at all and rather a person. A Builder AI situation Really hoping it's option 1 because that's pretty smart. Feels like that scene from Silicon Valley with the ...

oh the irony - this is literally deception

hmmm if grok is maximally truth seeking, seems like OAI is taking the stance of maximally helpful?

GPT5 live tweet 🧵