Carlos Donderis 🌊

The Definitive Guide to Harness Engineering

@mopx 😏

Extraño trabajar full time con Ruby on Rails. JavaScript was a mistake 😂

BTW I knew this was going to happen. cads-tech.dev/artificial-dum…

Tips to get more out of Opus 4.7 by Boris itself :)

Artificial (dumb?) intelligence cads-tech.dev/artificial-dum…

Es una locura. Estamos todos creando un vocabulario esperando entendernos pero hablando de cosas distintas. No ayuda tampoco que todo cambie y evolucione cada mes, y que estemos bombardeados por contenido generado por creadores del mismo buscando y mezclando buzzwords por doquier

Why Your “AI-First” Strategy Is Probably Wrong

I have been a GM running D&D Campaigns for decades. I can't explain how good Claude is as a GM assistant. One day I should explain my workflow :D Still very HITL but quite powerful

Scaling Managed Agents: Decoupling the brain from the hands anthropic.com/engineering/ma…

How We Built a System for AI Agents to Ship Real Code Across 75+ Repos mabl.com/blog/how-we-bu…

Agentic AI Engineering Workflows 2026 sesamedisk.com/agentic-ai-eng…

Background agents work across the entire SDLC. background-agents.com

La vez que Felix Baumgartner saltó a la tierra desde la estratosfera. Estuve todo el video con el culo en las manos.

AI is everywhere. The agentic organization isn’t—yet mckinsey.com/capabilities/p…

Protecting Our Systems with Intelligence engineering.block.xyz/blog/protectin…

Gemma 4

I liked this article from Block. Very Jack Dorsey style. Flattening organizations, evolving classic rigid roles and leveraging AI as the brain to fuel your organization is a good premise. Block shared more insights about how are they going about it than many (myself included). I loved the idea of the World Model aplied to an organization and I will likely use it for myself. But getting the world model to properly provide accurate information about decisions and priorities in a large and active organization is far from a trivial challenge, and I think, as of today, technology is not mature enough. But I'm sure it will be soon enough. block.xyz/inside/from-hi…

Genuinely didn't expect this. Left @karpathy's autoresearch running on a Mac Mini over the weekend. 259 experiments, no intervention. It landed at 1.353 val_bpb — a 30% improvement from where it started. For reference, the Mac Studio (4x the memory, 4x the price) took 5 hours of guided work to reach 1.29. The Mini got within 5% on its own. It just needed time. The weird part: it kept making the model smaller. Every improvement it found was about speed — fewer layers, smaller batches, tighter attention. More optimizer steps in the same time budget. On Apple Silicon, throughput beats scale. That wasn't obvious to me. tiny-lab is the control plane I'm building around this. Auto-restart, eval harness, experiment ledger, promotion protocol. Open source. github.com/trevin-creator…

On tokens and meaning cads-tech.dev/tokens-and-mea…

Don't need validation from Press Releases. We know the painpoints we are finding through hard work and experimentation. Double checking with leads in the industry is important to avoid ecochamber factors (actually this is super important) but don't delay action until validating is key.

@CaDs 🥹 Thanks for all the support!

Codebase standardization and structuring is essential with agents, which is exactly what my org is focusing on here: "Agents are most effective in environments with strict boundaries and predictable structure, so we built the application around a rigid architectural model."