Carter Sullivan

[Carter Sullivan @cesullivan99]

@RapidResponse47 @POTUS @NICKIMINAJ Show me a random clip of Trump and ask me if it's a deepfake. Guessing based on content of the speech alone (ie not processing the video), I've got no chance 🤣

[Carter Sullivan @cesullivan99]

@handotdev And same can be said for other successful founders/CEOs (ie Alex Karp) who go viral by promoting narratives that are certainly exaggerated/embellished in comparison to reality - it’s just what has to be done for good PR in this media environment.

[Carter Sullivan @cesullivan99]

@handotdev Assuming this is just rage/engagement bait. If it isn’t, it’s an interesting example of founders taking successful founders’ PR strategies (ie Travis wants to build hype/aura/a narrative around atoms, goes viral talking about the grind) quite literally

[Carter Sullivan @cesullivan99]

SF gas pumps know wassup

[Carter Sullivan @cesullivan99]

Challenge: listen to this on a run and don’t smash every PR

https://t.co/4OEDPz9QM2

[Carter Sullivan @cesullivan99]

@camiinthisthang That said, I do think that ultimately a lot of architecture and system design is just pattern recognition (exhibit A: prepping for a system design interview), and with the correct context and harnesses, models will eventually do a great job at this

[Carter Sullivan @cesullivan99]

@camiinthisthang Bad architectural decisions can compound in ways that wreck a system, and preventing them from happening early is critical. Even if there’s just one architect who approves high-level architectural decisions, it will make a ton of business sense

[Carter Sullivan @cesullivan99]

@signulll @signulll just curious, why say the user expectations have gone up? Feels like pre-slop era user expectations are finally getting applied to sloppified apps; the grace period for accepting shitty AI applications, simply bc they were built with AI and it was novel, are over.

[Carter Sullivan @cesullivan99]

@paulg Maybe in the pre-slop days where “shipping new stuff” meant a human actually thought through the feature and built it with intention. Would be more skeptical of this claim in a post-slop world

[Carter Sullivan @cesullivan99]

@Polymarket Uhhh ok. But could Opus 4.7, GPT 5.5, (or even older models!) etc, have done the same w the right harness and steering? My guess is yep. Things like this just feed the “Mythos = end of the cyber-world as we know it” narrative

[will o’brien @Willob]

Lol same guy btw

[Carter Sullivan @cesullivan99]

@danywander Ah, yes. OpenAI, Anthropic, DeepMind, and…Salesforce 🤣

[Carter Sullivan @cesullivan99]

@naval Free markets != other people? 🤔

[Carter Sullivan @cesullivan99]

@alliance @freeportmarkets 🚀

[Carter Sullivan @cesullivan99]

@DavidSacks @nlw By the end of June, Anthropic patches this jailbreak, administration lifts the ban, and even more people use Anthropic than ever before.

[Carter Sullivan @cesullivan99]

@signulll Regardless, this is one of the most fascinating questions rn - for what use cases will locally run models suffice, rendering closed source model providers unnecessary? And for which cases will closed source reign supreme

[Carter Sullivan @cesullivan99]

@signulll What type of products? ie, chat based vs agentic coding tools, and purely b2c or also b2b? Open source?

[Carter Sullivan @cesullivan99]

@AnthropicAI @trq212 This is the best marketing Anthropic could ever ask for