AI Engineering Digest

Short commentary episodes that track the weekday written briefing. Fraser hosts solo — opinionated takes on what changed for people who ship with models, with a light pass on local AI and Apple Silicon.

Subscribe via podcast apps using the RSS feed, or play episodes here.

Issue 25 5:18

Issue 25 — Ollama 0.40 RC makes MLX the default on Apple Silicon

Fraser on Ollama v0.40.0-rc0 making MLX the default on Apple Silicon (RC sandbox only; keep production pinned on 0.34.4), mlx-vlm 0.7.3, Gemini 3.8 Live with Live Avatar GA, the Australian PM's Medicare portal disclosure involving an OpenAI agent, Anthropic billing pre-output refusals, Databricks Unity Gateway CLI, and TRL 1.14.0.

  1. Issue 25 Ollama 0.40 RC makes MLX the default on Apple Silicon Fri 25 Sep 2026 · 5:18
  2. Issue 24 llama.cpp v0.5.0 Thu 24 Sep 2026 · 3:40

Earlier draft audio is not listed here yet. New episodes follow each weekday issue after publish approval.