Aprendizaje IA de la Semana: 27 de abril de 2026

Aprendizaje semanal de IA

🛠️

Herramientas / productividad

ryoppippi/ccusage: A CLI tool for analyzing Claude Code/Codex CLI usage from local JSONL files.

CLI tool that reads local JSONL logs from Claude Code or Codex to give fast, detailed usage reports. It supports daily, monthly, session‑based views, per‑model cost breakdowns, timezone and locale options, and even an MCP server to expose the data to Claude Desktop. Installation is optional; you can run it directly with npx ccusage@latest.

Devin for Terminal

Devin now runs as a local coding agent inside your terminal, giving you full access to your codebase, tools and environment while letting you hand off work to the cloud when needed. Built in Rust for speed, it works even on a VT‑100 and supports frontier models such as Opus 4.7, GPT‑5.5 and the proprietary SWE‑1.6. The project promises a seamless split‑screen workflow between laptop and cloud.

Arvid Global

Arvid is an AI‑powered platform that filters and analyzes public procurement notices from Colombia’s SECOP system. It shows only the opportunities that match your profile and provides a win‑probability score to help you focus on the contracts you are most likely to win. The tool aims to reduce noise and increase efficiency in bidding processes.

steipete/wacli: WhatsApp CLI

wacli is a scriptable WhatsApp client built on the whatsmeow library that pairs as a linked WhatsApp Web device. It mirrors your message history into a local SQLite store with FTS5 search, lets you send text, files, reactions and manage contacts/chats/groups from the command line. Output can be requested as JSON for automation, and the tool includes diagnostics, backup and media handling features.

🤖

Agentes / asistentes

ora // Agent Readiness Ranking

ora provides a ranking that evaluates the readiness and capabilities of different AI agents. The leaderboard helps users compare agents on various dimensions, offering a quick way to see which models are performing best in agent‑centric tasks. It is maintained by era labs and focuses on the evolving landscape of autonomous agents.

🧠

Modelos

Carlos Santana en X: “Esto es interesante: los modelos de Anthropic sufren más el “hablar” otros idiomas a nivel de uso de tokens.”

The tweet notes that Anthropic’s models consume substantially more tokens when generating text in Spanish compared to English. Specifically, Spanish usage is about 1.62 × the token count of OpenAI’s English baseline, highlighting language‑dependent efficiency differences between model families.

⚡

Optimización de flujos

Atai Barkai en X: “UI is Dead. Long Live Generative UI”

The thread argues that two parallel trends are reshaping software: the “headless web,” where agents interact solely via APIs/CLIs, and “generative UI,” where agents create dynamic interfaces for humans in real time. Examples from Salesforce, Slack and Anthropic show both approaches co‑existing, suggesting the interface is splitting rather than disappearing. The piece encourages developers to give agents a real UI instead of relying on text‑only replies.

📚

Gestión de conocimiento

[1hr Talk] Intro to Large Language Models - YouTube](https://www.youtube.com/watch?v=zjkBMFhNj_g)

A one‑hour YouTube presentation that introduces the fundamentals of large language models, covering their architecture, training processes, capabilities and common applications. The talk is aimed at viewers who want a concise yet solid overview of how LLMs work and what they can be used for.

🔗

Otros enlaces de interés

AI is frying our brains — here’s what leaders need to do about It | Fortune

El artículo de Fortune resume investigaciones que muestran que, contrario a las primeras promesas, la IA puede aumentar la carga cognitiva y llevar al agotamiento. Explica cómo liberar a los trabajadores de tareas de bajo nivel los empuja hacia trabajo de alto nivel más exigente, mientras que el entusiasmo por las nuevas herramientas puede enmascarar el costo mental real, sugiriendo que los líderes necesitan equilibrar la adopción de IA con el bienestar de los empleados.