Matthew Green

@alex_whedon Well, this is cool

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.

Have I been building an AI GTM engineer without knowing it? Well, maybe. But only the enrichment, lead finding, and outreach parts for the highest converting subset of your SOM. But what use is that? 😉

Stop asking for 'a few minutes to chat'. Start asking for specific introductions to solve specific problems. Be precise with your asks.

In case you didn’t get it, Chasqui is the map.

Your network is your most underutilized asset. You already know the people who know your next big client. You just need the map to find them.

@eriktorenberg I mean, sounds like you’re looking for a podcaster

I’m hiring someone to convene quality conversations with experts across sectors in group chats & IRL. They need to be online enough to know the memes/discourse but also functional enough to operate professionally at a high level. It’s surprisingly hard to find someone who is both

@Kekius_Sage Well, “nothing” can surpass light speed. But “something” can’t. Need negative mass things to do anything with it.

Are we 100% sure nothing can surpass light speed?

@i_am_brennan That’s mental

1.1B tokens today and about 4 symphony/codex tasks in the queue. Very hard to do with GPT 5.5. Could do this in a half day in 5.4. Tokenmaxxing is the way

@levelsio g.co/gemini/share/7…

@levelsio Thinking you build a two way valve to let the window “breath” while forcing it through a maze like insulated muffler

I still haven't solved the CO2 bedroom challenge You open the window and you wake up from a 6am garbage truck or barking dogs and sunlight You close it, you suffocate in 1200 ppl at 5am I guess you really need some mini tube in your wall with a vent that opens and closed based on internal CO2 but how do I build that?

Got it, they’re trying to sell me things

Why is my twitter feed entirely “what is everyone building, comment below” posts. Engagement?

@bscholl If you chase two rabbits, you will lose them both

We need alt phrases for “kill two birds with one stone” and “more than one way to skin a cat” that don’t involve being cruel to animals

Having a real tough debate re: credit card up front. I need to balance: - upfront enrichment costs - good initial user experience (first query) - Funnel size and not losing users right at onboarding when I ask for cc I think this is a take 2/3 situation. At least for now.

Cold outreach is a numbers game. Warm intros are a trust game. Which one do you think closes faster? #Founders #Sales

@AishwaryaDevv So really simple fix I’ve found, create an OVERVIEW.md with: -project structure -intent/goals -to dos -context -components Then tell whatever tool you’re using to update it whenever you finish a task. And look for it and read it at the start of each session

Most annoying part of vibe coding? Re-explaining the entire context to your IDE after switching chats

@rcmisk This is why warm intros win. Getting those first 100 customers is the real game

unpopular opinion: your product is the easy part now. AI writes code. no-code ships fast. anyone can clone your idea in a weekend. the only thing that's actually hard to copy is an audience that trusts you. are you building that, or just building features?

@mscode07 Chasqui.ai turn your connections into customers

Drop your Product Let's do some Marketing!!

@yazinhal That’s awesome

I made the Alarmy for screen time. Complete high-friction challenges to access your apps. And yes... you can even gamble on your screen time. Link below :)

@aryanlabde Distribution is the real moat now. Anyone can build

Worst part about building solo: you ship something genuinely useful and nobody sees it.

So Claude design is pretty cool. Turned my website into a decent pitch deck. I see the vision, it’s my own Shadcn component library.