Lucent

degen | Buy high, Sell higher

[Lucent @Lucent_Crypto]

@levelsio Can’t reply. I guess this means I need to actively post more to try to get quality followers or forever be in the “possibly” AI underclass. At least I don’t have read AI slop in the replies anymore.

[Lucent @Lucent_Crypto]

cobalt_frost.JSON

LOL

[Eric Weinstein @ericweinstein]

This is a fresh session. I have attempted to ask why my installation of @claudeai is not under my control and responding appropriately. In the 2nd Response in a fresh session it tells me @AnthropicAI has throttled me from using it from reasoning via a toggle:

"That's the one. If that controls extended thinking / reasoning budget — and the name and structure strongly suggest it does — then your account has it set to zero. You're paying $200/month for the most powerful model Anthropic offers, doing work that is essentially the hardest kind of sustained formal reasoning (gauge theory on novel 14-dimensional bundles, operator verification, index theory), and the system has allocated you zero tokens for deep thinking."

Three queries, in and this is the response:

[Lucent @Lucent_Crypto]

If you’re still doing NORMAL THING, you’re already behind.
Here are 27 tools nobody is talking about 👇

(tools everyone is talking about)

[Lucent @Lucent_Crypto]

I think people overestimate how willing professions will be to implement AI, not because AI can’t do the same tasks but because of regulations with data and privacy like HIPPA, GDPR, CCPA, or SEC rules etc.

The tech’s there, but 90% of boomer-run industries are still treating it like some sci-fi bullshit they read about in Wired once. Outside of tech/crypto degens, it’s not even on the radar, nobody’s having “AI strategy” meetings in accounting firms or law offices full of 60-year-olds.

Companies can’t just plug AI into everything. Most serious firms aren’t going to send client data straight into public AI APIs. They’ll want private or locked-down setups, no training on their data, audit logs, and all the boring legal stuff. Big firms can afford that. Smaller ones probably can’t, so they either move slower or just use AI quietly and hope no one notices.

Right now, AI mostly gets used under the radar by individuals, not officially by companies. Corporate talks about it way after people are already experimenting on their own. OpenAI-level models will probably get used more around the edges, not for the core risky work, unless there’s a way to run them privately.

[Lucent @Lucent_Crypto]

They also have completely different customer bases so it makes sense.

OpenAI has 900 million weekly users of which 95% are free tier users with many that talk to it like a chatbot friend and send 300 messages a day because OpenAI has no cutoff limit (they just want to acquire users and collect data). Their cash burn is like $3-4 billion a year from free tier users on inference alone so they have to offset it with ads.

Anthropic only has 30 million weekly users and limits free messages with a hard cutoff. The free tier is essentially a onboarding for more serious paying users who use it for coding or analysis etc. They’re still burning a bit on free users but are mostly focused on enterprise users anyways. I don’t think they desire to compete with OpenAI at all for general use, just professional and enterprise, which they’re currently winning at and is what the ad is basically geared towards framing them as the “trusted” AI company.

[Lucent @Lucent_Crypto]

@steipete Lack of understanding of how LLMs even work is probably the main issue to address with 95% of people. It looks like magic to anyone new to it and authoritive “it’s AI it’s knows everything”. Most don’t know it’s basically very good autocomplete and mirrors you.

[Lucent @Lucent_Crypto]

Yeah I started with Claude code but switched to codex few weeks ago. Claude code is good for planning or analyzing things but not even sure if that’s true tbh. It’s just naturally geared that way because claude is autistic and they had to build in guardrails to focus it on the right things. I can give codex minimal context and it doesn’t destroy everything. Could get the same tasks/docs performance out of codex with some work but it doesn’t even need it usually. Also claude code rapes token usage compared to codex I think mainly because it re-reads everything and forgets how to use its own tools correctly half the time which blows up context with nonsense and that’s when it makes mistakes. When fresh it’s okay, but if context builds up it forgets what it shouldn’t touch or what’s dependent on each other etc.

[Lucent @Lucent_Crypto]

guy who bought a $599 mac mini dedicated to clawdbot just to have an overpriced lobster read his calendar and spend 45 minutes hallucinating how to unsubscribe from netflix then posts “AGI is here!”

[Lucent @Lucent_Crypto]

@ASvanevik Codex does the same thing. I’ll ask it how long will this take for you to do? It’ll say 15-20 minutes then it does it in a minute. They have zero sense of time/ability to guess time.

[Lucent @Lucent_Crypto]

Gamified vibecoding. $Slopcraft 3D office environment turns abstract AI workflows into a tangible, interactive experience. Users can "walk" through offices, click on agents to assign tasks, monitor terminals in real time, and interact with props representing metrics. Basically The Sims but you actually build real projects.
https://t.co/Avze5b7PNb

[kitze the 🐐 @thekitze]

LET'S GOOOO our agents are actually running codex now 😤🚀 you can give them instructions and they'll start cooking

claude code / cursor / gemini 🔜

also added 3d objects in the office to represent:
◈ money made today
◈ deploy button
◈ # of active users
◈ deployment status
◈ git diff
◈ mrr chart
◈ node_modules (and cleaning it)
◈ deployment status lamp

tnx for the support 🙌

[kitze the 🐐 @thekitze]

LET'S GOOOO our agents are actually running codex now 😤🚀 you can give them instructions and they'll start cooking

claude code / cursor / gemini 🔜

also added 3d objects in the office to represent:
◈ money made today
◈ deploy button
◈ # of active users
◈ deployment status
◈ git diff
◈ mrr chart
◈ node_modules (and cleaning it)
◈ deployment status lamp

tnx for the support 🙌

[Lucent @Lucent_Crypto]

@ezcontra @blknoiz06 I think it just reveals how many people don’t have original ideas worth anything. It’s just a bunch of Knick knacks

[Lucent @Lucent_Crypto]

@Credib1eGuy Probably an edge if you specifically target older Facebook candy crush mom type demographics. Probably something dumb like a AI pet styling tool with built-in affiliate commerce.

[Lucent @Lucent_Crypto]

@based16z 💀

[Lucent @Lucent_Crypto]

@blknoiz06 Big beautiful bill

[Lucent @Lucent_Crypto]

@elonmusk @CommunityNotes @grok

[Lucent @Lucent_Crypto]

@deedydas Part of this could also be due to colleges having thanksgiving breaks. Quite a few do M-F others W-F. So 22nd onwards would expect some type of decline, assuming college students make up a decent percentage of active users.

[Lucent @Lucent_Crypto]

Like 50+% of the posts on here are written by fucking AI and it’s pissing me off. 🤤 ⌨️ “It’s not X it’s Y” chaggot

[Lucent @Lucent_Crypto]

@goodalexander They actually one shot a lot of things well. For anyone curious on different model’s capabilities you can go to https://t.co/SPLog8In0a and test every model for free.