Opiniones personales que no duran mucho.

Chile
Joined January 2009
This keeps happening to Tether because of a structural need I’ve mentioned a few times. To run business, you need two banks: one to custody the money mountain, and one to do the high-volume incoming and outgoing transactions. Tether has one reliable bank-like entity (Cantor).
Exclusive: Tether has some funds stuck at one of its offshore banking partners, which is facing liquidation risks after U.S. seizure of assets. Tether said its assets held at the bank, EQIBank, are limited and represent less than 0.034% of its assets. Story with @MichaelRoddan: theinformation.com/articles/…
10
16
4
263
39,714
Leo Soto retweeted
Prediction time. @dhh just asked over a thousand attendees at #railsworld how many people still write code by hand on a regular basis. Less than 10 people raised their hand. Let that sink in. Next year the question will be how many people are still doing manual PR code reviews or directly looking at code on any regular basis. Result will be similar.
47
68
14
630
69,772
Leo Soto retweeted
Being a computer scientist who refuses to find anything about LLMs interesting right now is a bit like being a geneticist who refuses to find anything interesting about the recently opened Jurassic Park
228
185
44
3,942
164,878
I just heard: a large company is moving from GChat (Google's Chat product) to Slack, because... Slack has much better agent operability (true) Who thought AI+agents would be a big selling point for Slack? Or that Google would sleep on agents + GChat!
112
25
16
1,138
307,981
Leo Soto retweeted
enterprise sales is nothing like sales … it’s 1:1 consulting to land and expand $MM deals
The full enterprise sales cycle, step by step with @jjen_abel Most people think there are 5 sales stages. There are actually 15. Skip a step and 💀 We discuss: 🔸 The “pincer model” for landing the first meeting 🔸 How to craft a winning 2-3 sentence cold outreach pitch 🔸 How to run an intro call that extracts maximum intelligence 🔸 The correct 2-3 day pilot structure 🔸 Pro tips for navigating pricing and procurement 🔸 So much more 84 minutes of enterprise sales alpha. Listen now 👇 youtube.com/watch?v=YS9In813…
19
18
3
399
51,058
Leo Soto retweeted
🧩 DeepSeek Harness v0.1 is now available in Developer Preview! 🔹 We’re opening it up to developers building agent harnesses worldwide and open-sourcing the codebase in MIT license. 🔹 Powered by the Cordis meta-framework, DeepSeek Harness is an agent harness built around one core idea: Everything is a plugin. Models, tools, skills, sessions, sandboxes, filesystems, loops, orchestration, and UI are ALL implemented as plugins, and can be mixed, matched, replaced, and extended. Try it now! github.com/deepseek-ai/deeps…
847
2,444
1,101
20,068
4,499,822
We can finally talk about it: We found a way to extract hidden reasoning of frontier models using a vulnerability in the APIs of every frontier AI company. We verified that our reasoning token count matches billed API thinking tokens 1:1 for most of the prompts we queried.
380
1,960
620
13,512
3,539,688
Leo Soto retweeted
It's time to reclassify PDFs as boomer technology I just want to be able to read an article on my phone!
102
51
15
1,369
93,061
Leo Soto retweeted
Replying to @mitsuhiko
I'm terrified that some day a ML algorithm somewhere will decide to cancel my Google or GitHub or iCloud account and my 20+ year digital life will blink out of existence
10
4
1
104
7,229
Leo Soto retweeted
I still sometimes see people saying "if you know how to write the code, it's faster to write it yourself" I'd argue the exact opposite: if you know how to write it, you gain nothing from doing the typing yourself - outsource that to a coding agent!
173
72
16
1,479
100,323
Attempted to get USD from a Japanese debit card at a U.S. ATM (yeah, weird even for me) and was presented with a screen clearly designed to bamboozle a tourist into paying 13.5%, and thus will repeat the industry advice: if a machine asks you if it can convert currency, answer No
35
90
14
4,767
342,502
Leo Soto retweeted
Agentic programming is just micro/macro from RTS games Even if you are 99th percentile at micro, a player who has better macro than you will beat you purely by economy and having a bigger army So focus on macro and getting the most productivity out of multiple agents doing work concurrently Then for the truly difficult tasks that require injecting supreme engineering taste and decision making, you can micro agents and direct them When two players have similar macro ability, the one with better micro will win by squeezing extra few percentage points in efficiency
being in the loop is important but i keep seeing people misunderstand this - your job isn't to micromanage the agent and get it to execute its steps in a way you think is better than what it'd do on its own it's to come up with cool things for it to work on
13
11
2
284
21,519
Strikingly similar to product done right (focus on the concepts allowing you to solve the underlying hard problem) vs the feature factory (focus on producing the marginal feature at the superficial layer).
One thing that is not obviously present in the "don't look at the code" debate is that two AI-coded codebases can be reached in different ways: 1. controlling all the main ideas, so you have a mental big picture of what is going on. 2. with brute-forced prompts to get results.
1
192
The whole point is to not align to the 24hr window. Temporal load balancing.
Why are the limits every 5 hours, surely 6 hours would make more sense? 4 equal chunks within 24 hours
5
3
1
309
40,231
Leo Soto retweeted
It is my belief that many devs right now are not maximizing what they can do with automatic programming because they still look at the code. Doing it makes you the bottleneck. Your time is better invested in new ideas, QA, design, and asking yourself what is your goal.
271
255
130
2,713
506,032
Leo Soto retweeted
The idea of "AI employees" feels so short-sighted to me - both disrespectful to humans and a complete misunderstanding of what these tools can do and how to best put them to work You may as well start adding Excel spreadsheets to your org chart
178
184
54
1,747
126,738
Leo Soto retweeted
This is why the BEAM is the perfect substrate for agents. Each one runs in an isolated, supervised process with a GC-enforced memory cap and a wall-clock kill. A runaway loop kills one struct, not your node. 100k concurrent agents is a struct each, not a fleet of VMs.
2
3
23
2,262
¿Cuándo desaceleró la economía chilena? Mediados de 2012 Por ser útil para el debate actual, vuelvo a postear un documento que preparamos hace dos años publicado en nota de ICARE-UAI icare.cl/assets/uploads/2024…
3
30
1
71
6,686