Blog posts in English
I write my posts in Portuguese first, for a Brazilian audience, and then translate them. These are all of them in English, newest first. The original Portuguese blog has the same posts.
The Era of Experience for AIs
And the end of the Era of Human Data?
OpenAI's Guide to AI Agents
OpenAI released a practical guide to building AI agents: when it's worth it, orchestration patterns and guardrails, plus LangChain's pushback.
Between Amazement and Indifference: The Age of Accelerated AIs
The innovation paradox: why does the "new" get old so fast?
OpenAI's new o3 model playing GeoGuessr
I played GeoGuessr with OpenAI's new o3 model, and watching its chain of reasoning was the most impressive part.
Agent2Agent Protocol (A2A) - Google's AI Agent Protocol
Google just made another important move on the AI AGENTS chessboard: the Agent2Agent Protocol (A2A), and how it complements Anthropic's MCP.
EVERYTHING about AI Agents
Paper 'Advances and Challenges in Foundation Agents'
Language Models (LLMs) Have Officially Passed the Classic Turing Test (?!)
A UC San Diego study found that GPT-4.5, given the right persona, was picked as the human 73% of the time in a classic three-party Turing test.
New Yann LeCun paper - Transformers without Normalization
A new paper by Yann LeCun proposes replacing normalization layers in Transformers with a single, surprisingly simple operation called Dynamic Tanh (DyT).
The Multimodal Revolution in AI Image Creation (and Its Effects on the Job Market)
Google and OpenAI's new multimodal models don't just understand text, they see and create images directly. Here's how it works and what it means.
Anthropic's Model Context Protocol (MCP): The Complete Guide
Anthropic quietly released MCP in November 2024, an open standard to connect AI assistants to data sources. Here's what it is, how it works and how people are using it.