Looks great in my app too. I did notice your video James specifically had some frame drops which I didn’t notice with Sam’s or Adam’s (first one happens around 7 seconds in). Not hugely important for a podcast but might be something to look into. Great show, as always. Really nice to see @Adam Clark Curry on video

It's a piece of the puzzle, and it does support search (I had it recommend me 5 podcasts based on my OPML export), but you'll need an LLM somewhere in your stack that supports tool calling to use it.

It's also limited to 60 QPM (i.e., 1 QPS) and only free while testing, so I'm not sure anybody should be racing to bake it into a prod app. Still very cool though. Listen to the latest PC2.0 it's a good one.

Any time! It's pretty sweet that we can just download a dump of the DB. Thanks @Dave

Assuming Opus guided me correctly, I’m seeing 9278/422303 in the last 90 days

Based off querying https://public.podcastindex.org/podcastindex_feeds.db.tgz and I’m not sure how current that is
```
SELECT COUNT(*)
FROM podcasts
WHERE (url LIKE '%riverside.com%' OR url LIKE '%riverside.fm%')
AND dead = 0
AND newestItemPubdate > strftime('%s','now','-90 days');
```

This feed’s live item is incorrect. Not sure if it’s a wider issue or just this one.

I’ve been thinking a lot about PI search lately. I think there are a lot of improvements that could be made just to the API itself, even before AI is brought in. Elastic search would solve that `&` issue easily, and make things like “Accidental Tech Podcas” (1 letter missing) give results. It has that fuzzy search functionality built in and is very fast. Only thing is it might be cost-prohibitive, but i think self hosting options could get cheap.