Section
All published posts.
The text the Senate approved in December 2024 classifies AI systems by use, bans some, requires documentation and impact assessments for high-risk ones, and puts the ANPD in charge of coordinating oversight. In the Chamber of Deputies, the high-risk list, copyright, and the size of the law are still in dispute.
Pi Durable kept two autonomous maintainers working for four days, including through an upgrade of Pi itself. With no real demand, they wrote their own issues and merged 693 pull requests.
I read five postings open in September 2026 at Palantir, OpenAI, and Anthropic. They describe the same arc, from an open question to production adoption, and they ask for two deliverables: the customer's system and what it teaches the product back home.
CLM reads the situation and each option separately and picks the closest one. It is fast when the options repeat, and weak when the sentence needs nuance.
The tables make Jev and Laya winners on different slices. The choice depends on where labels are defined and how much labelled data already exists.
Laya is a non-autoregressive decision model under Apache 2.0. It trades the convenience of a hosted API for control over running, evaluating, and specializing the model.
Jev trades text generation for typed decisions with probabilities. The idea works when the answer space is closed, the question is narrow, and code stays in control.
Coding agents look more capable when every task starts from a clean checkout. New benchmarks show the cost that appears when patches, decisions, and technical debt carry into the next job.
Dockerfiles, pipelines, and Terraform are not just another diff. When AI generates code with access to the production path, review must follow the risk of the file.
After nearly 10 billion tokens in Codex, why GPT-5.6 Luna became my default model: cost, performance, context, and limits.