Libertando mentes e divulgando ideias.

Libertando mentes e divulgando ideias.

O debate que hoje incendeia o cenário político e as redes sociais sobre a abolição da escala 6x1 é, em sua raiz, uma imensa cortina de fumaça. Assistimos a um espetáculo de demagogia onde o progressismo moderno tenta, mais uma vez, dobrar a realidade material…

Bracket Creep: O Duplo Roubo Invisível do Estado Através da Inflação. Por que o governo lucra duas vezes com a moeda que ele próprio desvaloriza.

A historiografia oficial e o discurso político feminista contemporâneo costumam celebrar a entrada da mulher no mercado de trabalho assalariado como o ápice da emancipação feminina.

Libertando mentes e divulgando ideias.

Comprar Bitcoin sem KYC (Know Your Customer ou "conheça seu cliente") significa adquirir a criptomoeda sem enviar documentos de identidade, fotos ou comprovantes de residência para corretoras. Muito além da simples evasão de burocracia, essa prática é a espinha…

Learn how large language models use tokens, transformers and learned parameters, what inference does, and what an LLM cannot guarantee.

AI inference uses a trained model to answer new requests. See how it differs from training and fine-tuning, and where model serving happens.

Understand LLM context windows, durable memory, RAG and KV cache—and why advertised context length does not guarantee useful recall.

Compare retrieval-augmented generation and fine-tuning for private documents, changing knowledge, repeatable behaviour and Canadian data control.

An offline LLM can run without cloud access after preparation. Learn what must be downloaded, what still connects, and what air-gapped means.

Compare AI chatbots, fixed workflows and agents by autonomy, tools, risk and control—and learn when the simpler architecture is the better choice.

Open weights, open source AI and API access are different. Learn what each permits and what Canadian businesses must verify before deployment.

The tariff crisis is not an AI tariff. It is evidence that Canadian businesses need a tested alternative to U.S.-controlled AI platforms.

CUSMA remains legally in force, but recent U.S. actions weaken it as a planning assumption. Canadian AI systems need portability and fallback.

Canada can operate suitable open-weight models on Canadian-governed infrastructure without waiting for one domestic model to lead every benchmark.

D-Central’s position: Canada cannot leave vital AI inference dependent on a foreign administration and platforms it cannot govern or replace.

OpenAI’s Tibo joins me to break down what’s next for Codex, ultra-fast AI, recursive self-improvement, and the future of personal AI agents. We also get into OpenAI vs. Anthropic, why Codex usage is exploding, and how AI could soon work faster than humans…

Rustin talks about Scott Bessent: Is This Market Chaos By Design?...