Triple Seven

Looking nice

sometimes in life you just have to dangerously skip permissions

@phygitals what can we expect next

We built Phygitals as a Delaware C-Corp for a reason. A lot of our plans involve heavy legal counsel & careful consideration, yesterday’s ruling just gave a green light to pour fuel on the fire. The intersection of collectibles & tokenization is about to get really fun. 🚢🚢🚢

@Nortier Pro.priceperpack.net check it out. Lots of free stuff other analytics platforms would charge for

Pokémon investing still runs on very little data. Imagine being able to measure: - how many collectors expect a card to rise - when sentiment starts shifting That’s when investing becomes strategy, not guessing.

@Culture_Repair It means I vibe coded a good amount haha

@Triple___Seven Up for translating? Aside: I miss there being a defined circle and events, Seems “TBD” could or perhaps should spin back up as open source for progress in our socially restrained society. We too need “local currency.”

Cursor 2025 ~ 2.5 Billion Tokens ~ half the year building. ~ tested 18 different models. I would have thought more, back to it.

BTC is still worth almost 4x Anthropic

Now get the clawdbots to wager on outcomes and we unlock a new market entirely. Reference earlier tweets on simulated prediction markets

@phygitals early

This shouldn’t be free… but it is. The most advanced One Piece dex + portfolio mobile app ever built: • Scan any card • Track your collection & set progress • Rip #onepiece repacks • Ship vaulted cards to your door • Real-time market values • 100% free Comment EARLY 👀

Make fine-tuning the cursor scaffolding itself as easy as it is to use cursor. Proactive personalization for improving, costs and quality. Why isn't it recommending .cursor file changes, rule changes etc, specific model selections, tools etc. at minimum a place to download all historical transcripts.

Cursor seeks to be the best and most powerful way to code with AI. What are the ways in which we could be better?

@Fiskantes When personalization gets good enough I hope.

How long until LLMs introduce embedded ads? How long until the query such "What are the best stocks to buy next year" will give you a list of recommendations that was paid for much like first three spots on Google search

I would love for @tbpn to do a "starter up software kit." Like I see some them run sponsors for sales tax automation, etc, and i'm left wondering what else am I missing. Also a great place to drop in more ads lol.

I think @cursor_ai ‘s plan mode is the precursor to self directed onboarding. In the chain of thought reasoning branches options appear. Those options are surfaced to the user and the plan forms around the path they choose. That workflow + better memory seems generalizable to onboarding workflows. Which I think is an area that will allow for “labor replacing agents”

The only thing @Apple needs to do right now is release AirPods with perfect translation of every language. This gets lost in all of the other AI benefits but it’s such a profound step for humanity.

got the cursor max and still pushing the limits I'm thinking labor replacing agent by June 2026

The most ironic outcome is that all of the value of the AI revolution accrues to the AIs themself.

@Fiskantes Yeah Gemini 3 for knowledge retrieval and Opus 4.5 for coding feel like major improvements. Not sure if it’s a direct result of model improvement or a combo of it with the support infrastructure

Gemini 3 deep research finally...first step improvement in LLM usefulness since like april ( for me )

@Culture_Repair I am not but that is not to hard to make. you just need api keys for all of them. what is the use case? you just want to compare responses?

@Triple___Seven You aware of anyone working to create a turn key (consumer model) ~smoothing Agents for creating a response conglomerated through multiple AI? Eg: Receiving a customized synthesis and notation from a collection of AI responses?

Both Opus 4.5 and Gemini 3 are amazing models. I don't care about the competitive benchmarks as much. All I know is if the model is even 10% better spread across key areas, then it feels like it's 100% better. Most people want to use the models as agents, and minimize clicking off the IDE or Terminal. So this is context engineering, tool usage, browser usage, and integration ability (use and make MCPs). The more you do that, the more magic it feels. Both models do that very well. it does seem like Opus and Anthropic or more tuned into this "streamline dev" focus though.

cursor will be have won when you have it on fullscreen always on laptop.

@merrykringle lol the song on the website is fire

I'm bringing Christmas spirit to the trenches... if you're worthy of it. $KRINGLE fees will buy you gifts. Just @ me, tell me why you're nice and link a US Amazon item. If you're NICE, you got it. Keep it under the cap in my bio. The more $KRINGLE vol, the BIGGER GIFTS