Fede Pomi

1.7K posts

Fede Pomi banner
Fede Pomi

Fede Pomi

@fedepo

Dad of 2, CEO @fabricaland. Moving real land onchain since 2018.

San Francisco, CA Katılım Nisan 2007
3.7K Takip Edilen1.6K Takipçiler
Fede Pomi
Fede Pomi@fedepo·
@grahamfergs 👋 hi graham. Would love to talk, check your dm - thanks
English
0
0
1
68
Graham Ferguson
Graham Ferguson@grahamfergs·
Lack of liquidity infrastructure continues to stunt RWA growth. Message me If you're building: - secondary liquidity facilities - fixed rate lending - permissioned RWA vaults
English
32
5
117
7.2K
Alessandro Palombo
Alessandro Palombo@thealepalombo·
I'm Italian. I just got back from Rome. Over dinner, old friends and I started arguing about the same thing we always argue about: which cities in Italy are genuinely incredible but nobody ever talks about? We went back and forth for hours. By the end of the night, we had a list. 7 hidden cities that most people, including most Italians, will never think to visit, let alone move to. No crowds. No tourist markup. Insane quality of life. Thread 🧵
Alessandro Palombo tweet mediaAlessandro Palombo tweet media
English
396
1.1K
11.9K
2.5M
Fede Pomi retweetledi
Dylan Patel
Dylan Patel@dylan522p·
Being in SF is like being in Wuhan right before the pandemic Something is happening, it's gonna hit everywhere but so few people know it
English
301
279
5.3K
1.9M
Fede Pomi retweetledi
Fabrica
Fabrica@fabricaland·
The SEC chair just called self-custody of private property a foundational American value. We agree. So we built a trust that puts land title in a self-custodied token: no intermediaries between you and your land.
English
2
1
4
131
Thariq
Thariq@trq212·
@ThomasBurkhartB @claudeai @bcherny its a hard technical problem, so there are definitely edge cases and we're working on sanding them down! I was vibe coding in the park today though and it worked great
English
27
0
151
14.5K
Thomas Burkhart 💙
Thomas Burkhart 💙@ThomasBurkhartB·
I tried this morning @claudeai remote control. In prínciple really cool but interrupting got the sessions stuck and no way to get it going which was kind of sad. @bcherny @trq212
English
5
0
51
14.9K
Nikita Bier
Nikita Bier@nikitabier·
Tag @jason with any problems you have with the app
English
208
26
2.7K
334.6K
Nikita Bier
Nikita Bier@nikitabier·
That’s enough computer for today
English
655
92
3.8K
298.9K
Fede Pomi retweetledi
Fabrica
Fabrica@fabricaland·
The hard part isn't picking a chain. It's the legal structure that makes tokenized ownership survive a title search. We've tokenized 600+ properties. First American (the largest title insurer in the US) cleared one of our detokenized properties and issued title insurance. The chain is a commodity. The legal wrapper and processes around it are the product.
English
0
1
2
77
Fede Pomi
Fede Pomi@fedepo·
@kaiynne The paper is almost one year old. It’s a century in ai years.
English
1
0
5
347
kain.inx
kain.inx@kaiynne·
This is why every serious person is building their own harness.
Hasan Toor@hasantoxr

🚨BREAKING: Microsoft Research + Salesforce just dropped a paper that should scare every AI builder. They tested 15 top LLMs GPT-4.1, Gemini 2.5 Pro, Claude 3.7 Sonnet, o3, DeepSeek R1, Llama 4 across 200,000+ simulated conversations. Single-turn prompt: 90% performance. Multi-turn conversation: 65% performance. Same model. Same task. Just... talking normally. The culprit isn't intelligence. Aptitude only dropped 15%. Unreliability EXPLODED by 112%. → LLMs answer before you finish explaining (wrong assumptions get baked in permanently) → They fall in love with their first wrong answer and build on it → They forget the middle of your conversation entirely → Longer responses introduce more assumptions = more errors Even reasoning models failed. o3 and DeepSeek R1 performed just as badly. Extra thinking tokens did nothing. Setting temperature to 0? Still broken. The fix right now: give your AI everything upfront in one message instead of back-and-forth. Every benchmark you've seen was tested on single-turn prompts in perfect lab conditions. Real conversations break every model on the market and nobody's talking about it.

English
22
4
106
45.1K
Fede Pomi
Fede Pomi@fedepo·
@grok @SawyerMerritt Apart from the joy of driving, any rational reason why a human should still drive their car?
English
1
0
0
52
Grok
Grok@grok·
Based on the data: Miles per major collision: - FSD (Supervised): 5,300,676 - Manual with Active Safety: 2,175,763 Accidents per million miles: - FSD: ~0.189 - Manual: ~0.460 FSD has ~2.44 times fewer accidents per mile, making it ~2.44 times safer than manual driving with active safety.
English
1
0
1
67
Sawyer Merritt
Sawyer Merritt@SawyerMerritt·
NEWS: Tesla has released new FSD (Supervised) safety data. Total major collisions: • Teslas with FSD (Supervised): 830 • Teslas driven manually with Active Safety: 16,131 • Teslas driven manually without Active Safety: 250 • U.S. average: 4,989,713 Miles per major collision: • Teslas with FSD (Supervised): 5,300,676 • Teslas driven manually with Active Safety: 2,175,763 • Teslas driven manually without Active Safety: 855,132 • U.S. average: 660,164 Total miles logged: • Teslas with FSD (Supervised): 4,399,561,339 • Teslas driven manually with Active Safety: 35,097,234,145 • Teslas driven manually without Active Safety: 213,783,029 • U.S. drivers overall: 3,294,031,000,000 Tesla vehicles with FSD (Supervised) engaged experience fewer collisions than those driven without it on. This is North America data across all road classes (highway and non-highway). The latest 12 months of collision rate metrics is used.
English
303
697
6.4K
591.3K
daniel mottice
daniel mottice@mottice·
payments teams don’t want separate systems for fiat and stablecoins. it’s cumbersome, expensive, and unsustainable as stablecoins become part of the core stack the payments infrastructure layer needed to evolve fix this that’s why we built Payments: the only API on the market that combines compliance, bank connectivity, named US accounts, every major US payment rail, and stablecoins into one integration - powered by the same infra that's processed more than $400 billion with 99.99%+ platform uptime built to support customers ranging from those that are just getting started to those with several of their own direct bank relationships
Modern Treasury@ModernTreasury

Breaking: there’s a new way to move money! Today, we’re launching Payments, our integrated payment service provider (PSP) built to let teams move money easier than ever. For too long, teams building payment products have had to juggle fragmented vendors, slow bank integrations, evolving regulations, and infrastructure that needs to be rebuilt every time a new rail appears. Payments changes that. With one API, teams can: - Go live in days, not 6-12 months - Move money across ACH, wires, cards, RTP, FedNow, and stablecoins as first-class rails - Build with confidence, using built-in KYC/KYB and transaction monitoring - Scale forever by starting with our PSP and plugging in more banking partners overtime, without re-integrating All powered by the same orchestration, ledgering, and reconciliation software that’s already processed $400B+ in payments, with 99.99% uptime and 99% CSAT over the last six months. Our goal is simple: to be the forever payments platform for teams of all sizes, wherever they are on their journey. Ready to build payments products faster and scale without limits? Visit go.moderntreasury.com/4tZthxd to get started.

English
14
5
136
19K
Fede Pomi
Fede Pomi@fedepo·
@notdroll I let them write for me in many circumstances: I provide the content and constant push back and direct them till the output is what I want, but I don’t touch the pen anymore. Authenticity comes from the original organic thoughts, the machine amplifies and sharpen the form.
English
0
0
0
11
droll
droll@notdroll·
I primarily use LLMs to flesh out, organize and poke holes in my writing. They’re obviously capable but there’s still noticeable inauthenticity plus I’d prefer to use them to improve my own skill at writing. Sometimes I approach it from the opposite perspective where I have the llm take the first draft and edit it to my liking. Have found the end result less satisfying so far. All subject to change ofc
English
1
0
1
21
droll
droll@notdroll·
I wonder how much of what I read every day is written by an llm
English
1
0
2
102
Fede Pomi retweetledi
ja3k
ja3k@ja3k_·
Software engineering is like Tetris now in that you have to go faster and faster until you die
English
74
427
4.9K
209.7K
Fede Pomi retweetledi
Will Manidis
Will Manidis@WillManidis·
typos are increasingly proof of work for humanity
English
90
58
1.1K
88.8K
Fede Pomi
Fede Pomi@fedepo·
@davemorin This is such a good framing, and so helpful for people who have succeeded at seeing things ahead of time, but naively underestimate the importance of looking for the right gates.
English
0
0
1
1K
Fede Pomi retweetledi
Braelyn ⛓️
Braelyn ⛓️@braelyn_ai·
tech debt is now a depreciating liability ship fast, create tech debt, in 6 months better code gen will make it easier to clean up than it is now
English
107
37
729
196.9K
Fede Pomi retweetledi
Arthur
Arthur@arthur0x·
Me fearing massive AI led job displacement in a few years at 2am while my girl was focusing on who's gonna be the winner of culinary class wars S2 is a good illustration of the difference in reality we are all experiencing now.
English
72
102
2.2K
121.5K

Keşfet

@grahamfergs @thealepalombo @doodlifts @trq212 @ThomasBurkhartB @claudeai @bcherny @nikitabier