I'm still getting the habit of writing this small form content. It is very interesting because, to be honest, it looks to me like a small effort format that gives space between longer form content that requires more thinking and time in front of the machine.
The latest mastodon update brought quotes). I'm super happy to see mastodon community advancing in a healthy state.
Google releases MCP for all its services. I need some time to experiment with this but I hope I can do nasty things with drive.
Skills are now an open standard). It is cool to see some open standards to be proliferating on this new AI world. I am still concerned on how portable we should be from one LLM client to another (spoiler alert, it should be 100% portable with no friction).
Questionable ethics aside Anna's blog is always super interesting to read for technical details. This post on their efforts to backup Spotify is no exception.
I've been playing with text based adventures recently (blog post incoming), but it looks like there is smarter people than me doing it already.
Even if you don’t train models, walking through the llm evaluation guide will give you a lot of ideas. It is amazing that this kind of content is still free.