0xRutts

i’m an AI systems engineer who loves to research on ml models when not building secure and scalable AI systems

[0xRutts @0xRutts]

what happens to existing $200/mo subscribers mid cycle? will their usage net out at half the dollar in API spend compared to the old Pro $200 plan starting next cycle or from tomorrow?

[0xRutts @0xRutts]

@trq212 agreed! it can be a standalone skill if needed but usually models work well when we plan and execute on the go

[Thariq @trq212]

the right way to use model capabilities is not to ship 10x more features to prod

it's to spend more time understanding your users, trying experiments, building prototypes, learning about things you don't understand so that you can ship things that actually work

[0xRutts @0xRutts]

@thsottiaux would it be okay to switch my existing Astra medium session to Sol high (by sacrificing the cache once) ? Or better to continue in a new thread?

[kache @yacineMTB]

I can do incredible things with AI, but don't be fooled. It's because I understand how things work, and understand what's possible

You need to understand man

[María Benavente @merybenavente]

you know what else Jev helps classify?

people who actually understand ML, and people who don't.

[James Cowling @jamesacowling]

The real engineering starts *after* you’ve shipped

[DHH @dhh]

Agency is rarely given. It’s taken, then recognized.

[0xRutts @0xRutts]

true, but there are still a lot of orgs which don’t realize that either you need someone who knows how to drive AI or someone who is open to learn. otherwise, reckless prompting would result in a mess even an experienced engineer would hesitate to cleanup later on.

additionally, in a decent number of orgs I have seen that the early adopters are either classified as “vibe coders” or “slop shippers” and once they get the recognition that they were right , its too late and they are then left to clean up the actual slop which partial adopters created. and the only reward is more mess to clean nothing else!

[0xRutts @0xRutts]

@thsottiaux Plan mode? Astra and Sol are both good at planning and executing hand in hand as we interact with them. Plan mode sometimes leads to over-engineering just because model “feels” like the plan needs to be foolproof before execution even starts.

[0xRutts @0xRutts]

hey @thsottiaux @OpenAIDevs, i know lots of useful issues are coming in and the team might already be on this, but this one makes the codex app’s remote SSH workflow kinda unusable.

sending a message from the mac codex app adds a visualization sandbox root and breaks every shell command. i have to switch to the server terminal and send another message to get it working again.

added a repro, code refs, and an isolated test where reordering the mounts fixes the failure while keeping the tested protections:
https://t.co/0Ni4hKsNHq

[0xRutts @0xRutts]

astra doesn't need /goal

[Guillermo Rauch @rauchg]

Anyone that works in AI is working the hardest they've ever worked in their lives. On the surface it's somewhat ironic (AI should give us back time!), but the reality is that it's the most fun, fascinating and empowering epoch in human history. The intelligence revolution.

[0xRutts @0xRutts]

@Vai2052 Happy teacher’s day serr!!

[shafu @shafu0x]

fucking hilarious that AI is mandatory at work but banned in interviews

[ThePrimeagen @ThePrimeagen]

We got GPT 6 Before GTA 6

[0xRutts @0xRutts]

@thsottiaux stars and sparks?

[0xRutts @0xRutts]

@poteto perfect!! will try it out

[0xRutts @0xRutts]

@poteto poteto poteto

what if users could use their Cursor Pro / Pro+ / Ultra sub in the Grok Build TUI?
link Cursor in Grok Build, pick /fable-5, and it just burns Cursor “other models” limits?