Joaquin Trelleira

Founder of Uno Collective

[Joaquin Trelleira @joaquint]

AI is becoming a surprisingly good analogy for how people work inside organizations.

Context windows: Give someone too little context and the output won’t be good. But if they’re overloaded, their work will start getting worse and worse.

Model training: Someone trained to do one thing can do something similar, but not necessarily well.

Smarter models cost more, weaker ones are cheaper: hello, seniority. Any day now, Jr, Mid, and Sr will be replaced by Haiku, Sonnet, and Opus.

Agent orchestration and loops: One person alone can’t guarantee things will be done well. But if you organize people so they correct each other, with different layers of seniority, you’ll almost certainly get to something better.

Human-in-the-loop: More like CEO-in-the-loop. An organization runs on its own until something actually needs to escalate to the CEO.

[Alex Atallah @alexatallah]

We just announced our Fusion API:

- Fable-level performance on deep research tasks, at half the cost
- Better-than-SOTA performance using panels

The future of AI is neurodiversity, not single-model takeovers.

[Joaquin Trelleira @joaquint]

@HermesAgentTips And the templates?

[Joaquin Trelleira @joaquint]

I’m not really buying the whole “thousands of dollars in OpenAI or Anthropic credits” narrative.

Run Chinese models through OpenCode or Ollama and the difference isn’t all that big for the same cost.

[Joaquin Trelleira @joaquint]

@ThiagoMonechesi El creator marketplace de Meta te puede servir no solo por lo que convierta directo, sino que los videos en colab pautados desde tu cuenta probablemente puedan convertir mejor.

[Joaquin Trelleira @joaquint]

@ThiagoMonechesi Seguro te podemos ayudar

[Latinometrics @LatamData]

🇦🇷→🇺🇸 Argentina’s brain drain is quietly creating one of America’s highest-earning Hispanic communities.

Argentines and Chileans in the US now out-earn the average American household, while other Latino communities face very different economic realities. The reasons say a lot about migration, education, and opportunity.

🗞️ more here → https://t.co/al8n829aZW

[Joaquin Trelleira @joaquint]

@aronprins Thanks!

[Alexander Whedon @alex_whedon]

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.

[Joaquin Trelleira @joaquint]

@samgrows Hi! You mentioned that Claude can now “watch” videos. Could you explain the process a bit more? I couldn’t find that information on the website. There’s quite a difference between reading transcripts, taking a few screenshots, and actually watching the video. Thanks.

[Aran Komatsuzaki @arankomatsuzaki]

The non-English tax is real.

Sutton's Bitter Lesson, translated across languages and normalized to OpenAI English token count:

Hindi: OpenAI 1.37×, Anthropic 3.24×
Arabic: OpenAI 1.31×, Anthropic 2.86×
Chinese: OpenAI 1.15×, Anthropic 1.71×

Claude’s tokenizer charges a much higher linguistic tax.

[Ben Horwitz @horwitzben]

I made the anti-Grammarly. Mess up your emails with AI.

https://t.co/R6kdTpoc0W

[Joaquin Trelleira @joaquint]

@dotta Thanks!

[Joaquin Trelleira @joaquint]

@elonmusk Money problems and expectations for the middle class

[Claude @claudeai]

Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude.

Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day.

[Google @Google]

We’re introducing a new Search experience in @GoogleChrome that lets you open webpages side-by-side with AI Mode – no tab switching required.

Now, you’ll be able to compare details and ask follow-up questions while still maintaining the context of your search, whether you’re shopping for a new coffee maker or studying for a statistics midterm.

[Marc Benioff @Benioff]

Welcome Salesforce Headless 360: No Browser Required! Our API is the UI. Entire Salesforce & Agentforce & Slack platforms are now exposed as APIs, MCP, & CLI. All AI agents can access data, workflows, and tasks directly in Slack, Voice, or anywhere else with Salesforce Headless 360. Faster builds, agentic everything. 🚀
#Salesforce #Agentforce #AI
https://t.co/mxySdJS7HR