jonjon

Latam Lead at @Outlierventures

[jonjon @jonjonflo_]

@zachrose51 https://t.co/pLHQ4MMQ7J clients are big producers of electronic music events in LATAM

[jonjon @jonjonflo_]

@HannaSchiuma Interesting either way… Naval always wins

[jonjon @jonjonflo_]

@HannaSchiuma Several underlying fees though, no? It’s essentially a FoF + secondaries and spvs pluus angel list fees. It’s what you get for non-accredited 500$ minimum

[bebepuey.eth @bebepuey]

I've been saying this since last year: what's coming is the industrial revolution on steroids. This is the biggest shift we'll experience as a species. Not close to anything before.

And most people have zero clue about what just happened yesterday.

Last week I watched a talk by Nicholas Carlini, a research scientist at Anthropic. He pointed Claude Code at the Linux kernel source with a simple script and it found a vulnerability in the NFS driver that had been hiding since 2003. Twenty-three years of research and development, thousands of engineers, and nobody caught it. An AI found it in hours.

His words: "I've found more bugs in the last few weeks with Mythos than in the rest of my entire life combined."

That was last week. Yesterday it got way bigger.

Anthropic launched Project Glasswing. They built Claude Mythos, a model so good at finding software vulnerabilities that it autonomously discovered a 17-year-old remote code execution exploit in FreeBSD. One prompt, root access, game over. It already found thousands of zero-days across every major operating system and browser. Every. Single. One.

Now read this list of companies that immediately joined forces: AWS, Microsoft, Google, Apple, NVIDIA, Cisco, CrowdStrike, Broadcom, JPMorganChase, Palo Alto Networks, and the Linux Foundation.

That's not a press release. That's a war room.

AWS, Azure and Google Cloud together own 63% of the world's cloud infrastructure. The servers running your bank, your government, your hospital. When those three drop everything to work together with the company that built this model, it's because they understand what happens if this gets out without guardrails. States, banks, any institution that holds society together could be compromised in a single prompt. This is not a drill.

And here's the thing that should really hit you: probably less than 1% of the world's population will even hear about this. You're reading this because you're on X. That's it. That's the filter. The gap between what's actually happening in AI and what the average person knows is getting wider every single day. By the time mainstream media covers this properly, the world will have already moved three steps forward.

We're living through the biggest technological inflection point in human history and most people are scrolling past it.

Pay attention.

[jonjon @jonjonflo_]

@DiegoARRG DM muchas gracias!!

[moneyfetishist @moneyfetishist]

I am not going to motivate you because if you need motivation from a stranger on a plane the answer is stay

but I will give you the game theory

your corporate M&A gig is a repeated game with diminishing marginal returns. year 1 you learn everything. year 2 you refine it. year 3 you are executing pattern recognition. year 4+ you are being paid more to do the same thing with slightly larger numbers. the learning curve flattens but the golden handcuffs tighten because every year the comp goes up and the opportunity cost of leaving gets more painful on paper

this is a classic status quo bias trap. the payoff of staying is known and comfortable. the payoff of leaving is uncertain and scary. so you stay not because staying is optimal but because the asymmetry of regret is lopsided. you can imagine regretting the leap. you cannot as easily imagine regretting the years you stayed too long because that regret builds slowly and never hits you in one moment

here is where game theory actually helps:

in your M&A seat you are playing someone else's game. the firm sets the rules, the deal flow, the comp structure, the promotion timeline. you optimize within their framework. you are a very well-compensated player in a game you did not design. your upside is capped by whatever the partnership or MD economics look like. your downside is protected by a salary. that is the trade

owning a local business flips the entire payoff matrix. you design the game. you set the rules. the downside is real and unprotected but the upside is uncapped and compounds in ways a salary never does because you own the equity. a $2M EBITDA business bought at 4x and grown to $3M EBITDA over 3 years is worth $12-15M on exit. no M&A salary trajectory produces that kind of wealth creation in that timeframe unless you are a founding partner

the Nash equilibrium of your current situation: you and every other M&A professional are competing for the same promotions, same deal credit, same bonus pool. the competition is fierce because the players are identical. same schools, same skills, same hours. you are in a crowded equilibrium where everyone works 80 hours to stay in the same relative position

local business ownership is a different game with different players. the competition is a 62-year-old owner who stopped innovating in 2014 and a 35-year-old who inherited the business and does not want to be there. you walk in with financial sophistication, deal structuring experience, and the ability to read a balance sheet faster than anyone in the room. you are overqualified for the game which is exactly where you want to be. the best strategy in game theory is to play games where your existing skill set gives you an asymmetric advantage over the other players

the timing question is about optionality. every year you stay in M&A your financial optionality goes up slightly because you save more. but your operational optionality goes down because you get further from the reality of running anything. the M&A guy who leaves at 28 adapts to operations in 6 months. the one who leaves at 38 has a decade of habits built around delegating to analysts and reviewing decks, and managing a P&L feels foreign in a way it would not have 10 years earlier

but again. if you need me to motivate you, stay. the people who actually do this do not need motivation. they need a spreadsheet that shows the math works and then they cannot NOT do it. if you have the spreadsheet and you are still asking strangers for motivation the spreadsheet is not the problem

[jonjon @jonjonflo_]

@Simonpuebla_ @0xnico_ Good insight

[jonjon @jonjonflo_]

What does or not comes next but def sth to read. Be judicious

[Radigan Carter @radigancarter]

https://t.co/Ys7x2TMb2t

[bebepuey.eth @bebepuey]

My ADHD fighting my ADHD created this.

Feel free to collab =)

Claude Code - Monitorator -

https://t.co/iN8kge7EIS

[jonjon @jonjonflo_]

Mamma mia

[The White House @WhiteHouse]

STRIKE. 💥🦅

[jonjon @jonjonflo_]

@bebepuey @bombocommunity Way to go, vamoosss

[bebepuey.eth @bebepuey]

Starting Monday every @bombocommunity employee gets Claude Desktop + Cowork access, no matter the area. Plus a token budget to spend on whatever other AI platform they need.

IT and Product teams get Claude Code $200/mo each.

Weekly training included because honestly there's no way to keep up with this pace otherwise.

@bcherny @AnthropicAI companies need to start massive training programs on this. Now.

[Johan @Adityapandeydev]

Software engineering in 2026

[bauti.eth @BautiDeFi]

Not sure where I just saw this but this is a god send

EVM based skills for claudio, made by the one and only @austingriffith

https://t.co/LeFMx93HgB

[jonjon @jonjonflo_]

@adnaaan433 @Blayasensat

[Chris @ChrissGPT]

Anthropic had 16 AI agents build a C compiler from scratch. 100k lines, compiles the Linux kernel, $20k, 2 weeks.

To put that in perspective GCC took thousands of engineers over 37 years to build. (Granted from 1987 - however) One researcher and 16 AI agents just built a compiler that passes 99% of GCC's own torture test suite, compiles FFmpeg, Redis, PostgreSQL, QEMU and runs Doom.

They say they "(mostly) walked away." But that "mostly" is doing heavy lifting.

No human wrote code but the researcher constantly redesigned tests, built CI pipelines when agents broke each other's work, and created workarounds when all 16 agents got stuck on the same bug.

The human role didn't disappear. It shifted from writing code to engineering the environment that lets AI write code.

I don’t know how you could make the point AI is hitting a wall.

[jonjon @jonjonflo_]

@drewfallon12 FIN

[Drew Fallon @drewfallon12]

Introducing Fin: The world’s first AI Chief Financial Officer.

Fin outperforms humans 100% of the time.

RT + Comment “FIN” and I’ll send you an AI agent that saves 6-7 figures/year.

[litquidity @litcapital]

Excellent choice

[jonjon @jonjonflo_]

@BautiDeFi Si🤝