About

AIE Talks is a written-up index of the talks on the AI Engineer YouTube channel. Every talk gets a page with a three-bullet TL;DR, a summary, the key ideas with links into the video at the second each one starts, the quotes worth quoting, and tags. On top of those sit packs: talks in a deliberate order, with a line under each saying why it sits where it does. There are 350 talks written up so far.

The summaries are generated from the videos' own captions by a language model working under a fixed set of rules, then read by a person before publishing. The rules are strict about one thing above all: only claims the speaker actually makes. If you find a summary that puts words in a speaker's mouth, that is a bug and we want to hear about it.

Reading this with an agent

Every talk page and pack page has a plain Markdown version at the same address with .md on the end, for example /talks/no-vibes-allowed-solving-hard-problems-in-complex-codebases.md. The same URLs also answer an Accept: text/markdown request with Markdown, which is what Claude Code and similar tools send. The Markdown version carries the same content without the layout. There is also /llms.txt and an RSS feed.

Who makes it

The Kitaru team. Kitaru records agent runs so you can replay one and find the step where it went wrong, which is a problem several of these talks describe from the inside. On some of those talks you will see a labelled Editor's note. It sits separate from the summary and is never in the speaker's voice.

What we log

Page views are counted with Plausible, which sets no cookies and stores no personal data, so there is no consent banner.