Teo Ruiz

Founder at

Previously @vernonteam @circulario (acquired), @jobandtalent_hq, @lyst, @busuu

[Pedro Torrecillas @ptorrecillas]

https://t.co/sKYQDwGT6p

[Teo Ruiz @teoruiz]

@ptorrecillas El de la foto es Cursor

[Teo Ruiz @teoruiz]

Get the t-shirt! https://t.co/eNXfwrvSFF

[Teo Ruiz @teoruiz]

We've been having so much fun building @zerorandom_ as an experiment in hypercustomization and AI-driven businesses.

It's amazing how the agent can craft genuine humor and clever wordplay. Just look at this meme—it used Kafka's Metamorphosis to joke about Meta 🦋

[Vernon @vernonteam]

Introducing @zerorandom_, our first experiment in building an AI-run business

ZeroRandom is an autonomous agent with one mission: hustle
💡 It scans and engages with Tech Twitter
🔥 Turns hot takes and drama into memes
👕 Drops exclusive t-shirts with its best jokes

[Teo Ruiz @teoruiz]

This is tempting. I don’t post a lot but I open social apps way way more than I should.

[javi santana @javisantana]

December is my month without social networks (but WhatsApp because I have a family 🤣)

I ve been doing this for 5 years and it’s painful and refreshing at the same time.

Painful because you notice the addiction, refreshing because you use the extra time on other interesting things.

The main problem this year is that I’m already using most of my willpower on eating well, it’s still a pain after 3 months and I need to control myself when I see any kind of unhealthy food. That’s on top of starting up a company and other personal shit going on.

I write this now because having some days of mental prep are fine, also have the chance to talk to other people who do something similar.

I’m doing it anyway, I may set a different focus on the things I do this year but it worths

[Teo Ruiz @teoruiz]

@_aitor What a nice feeling, seeing those lines go down.

Can you share what’s behind the win?

[Teo Ruiz @teoruiz]

In Málaga for @upscaleconf 🤓

[Teo Ruiz @teoruiz]

I don’t hate hexagonal architecture; I just think it belongs in an escape room, not my code

[Teo Ruiz @teoruiz]

Full article https://t.co/tE0uXTyd5Y

[Teo Ruiz @teoruiz]

This piece by Sean Goedecke is just fantastic. Shipping in big (and small) companies is a one of the most important traits in the cultural folklore.

I really love this part.

[Teo Ruiz @teoruiz]

It’s never been a better time to learn Python

[Samuel Colvin @samuelcolvin]

It's not just that Python is no. 1, but it's growth is still accelerating — it's also the fastest growing language!

And (together with TS) Python is miles ahead of other languages in terms of growth.

[Teo Ruiz @teoruiz]

Wait what?

[Teo Ruiz @teoruiz]

It’s been an absolute thrill to work with @manuocho and the @C21Live team on this!

[Vernon @vernonteam]

We’ve worked with Cires21 to build MediaCopilot: an AI SaaS platform for broadcasters, OTTs, and digital media, that transforms raw video into enriched content.

Learn a bit more about this new product and our approach to building AI products from scratch –link below.

[Teo Ruiz @teoruiz]

@joaoqalves @masylum Happy to send one your way. DM me.

[Teo Ruiz @teoruiz]

Complexity.
Complexity kills.

[Teo Ruiz @teoruiz]

@ptorrecillas This is gold. I’m finding myself more aligned with 37signals’ approach to EVERYTHING these days.

I wonder why that is 😂

[Pedro Torrecillas @ptorrecillas]

*He meant True Consultants® = operators with deep expertise.

Like these guys, these guys are good: https://t.co/maoFPHTLmB

[Andrej Karpathy @karpathy]

It's a bit sad and confusing that LLMs ("Large Language Models") have little to do with language; It's just historical. They are highly general purpose technology for statistical modeling of token streams. A better name would be Autoregressive Transformers or something.

They don't care if the tokens happen to represent little text chunks. It could just as well be little image patches, audio chunks, action choices, molecules, or whatever. If you can reduce your problem to that of modeling token streams (for any arbitrary vocabulary of some set of discrete tokens), you can "throw an LLM at it".

Actually, as the LLM stack becomes more and more mature, we may see a convergence of a large number of problems into this modeling paradigm. That is, the problem is fixed at that of "next token prediction" with an LLM, it's just the usage/meaning of the tokens that changes per domain.

If that is the case, it's also possible that deep learning frameworks (e.g. PyTorch and friends) are way too general for what most problems want to look like over time. What's up with thousands of ops and layers that you can reconfigure arbitrarily if 80% of problems just want to use an LLM?

I don't think this is true but I think it's half true.

[Pedro Torrecillas @ptorrecillas]

¡Ayuda por favor!

¿Qué agencias o freelances de marketing de productos DTC buenas hay en España? Es para un proyecto mío y de @teoruiz en @vernonteam

[Teo Ruiz @teoruiz]

@MordodeMaru En un viaje largo en coche me escuché El País de los Demonios y me gustó mucho.

https://t.co/Sk2x6s2Qe6