I’m typically team @ianlandsman.com but not on this one.

Vectors created by an LLM will for sure capture the semantic similarity between a high heel and a shoe.

Put together a little demo showing pgvector working with Laravel Scout:

pgvector.laravel.cloud

I feel the same way. Lately I’ve been using both. Switch to Cursor for the AI generation and then review the changes and make manual tweaks in PHPStorm.

Everything about warzone is terrible except the game itself. Devs resting on their laurels.

Good looking scoop

Gotta be @astro.build. Though I really want to check out vapor.codes

Assuming you're referring to the pgvector package I created for Laravel Scout, I think that would pair well with @tjmiller.bsky.social's Prism to create a RAG workflow.

A Prism tool call that performs a pgvector search would give your LLM of choice access to your data for further processing.

Wait until you find out he bikes the continental divide from Canada to Mexico every year.

Good idea! I have two Apple Watches for the same reason.

bsky.app/profile/benb...

I guess it depends on what you do in post to get things sounding right. I just use a compressor and an expander so interfaces by Rode, MOTU, and RME work well for me.

If you do more something like the UA Apollo Solo could be a good choice since it can run plugins that can do almost anything.

Ended up taking 55 minutes.

Im sure it would help, but using an interface with built in DSP seems simpler.

Adding shadcn to the Inertia React variant would be 🔥

The “Show less like this” button really cleaned my feed.

When I signed up it was all anime and digital art by default and I don’t see any of that anymore.

I’d probably run the models with ollama and use Laravel to hit its localhost api.

github.com/ollama/ollam...

With the callback syntax you don’t have to use a controller

Always enjoy this, thanks!