I mainly use Selects for podcast assembly, and it's become the fastest part of my workflow. Long-form interview footage is the worst thing to log manually — you're scrubbing through an hour of two people talking to find the eight minutes that actually make the episode. Selects does that pass for me.

Since the MCP update I can drive it straight from Claude, and that made a noticeable difference in both speed and quality of results. I point it at the footage folder, it scans and probes everything, and I get a draft back before I've opened a single bin.

The biggest upgrade is that it can now actually see and analyse the video instead of only reading the transcript. I've tried transcript-only tools and they're fine for finding a quote, but useless for finding a moment. Now that Selects understands what's happening on screen, I can use it to cut highlight reels for events, where the good stuff is visual and nobody says anything that would show up in a transcript.

Handoff to Premiere is clean too — I get something I can open and start cutting, not just a list of timecodes to go re-find myself. Review collected by and hosted on G2.com.

My main request is support for stills as B-roll. A lot of my projects have a photo library alongside the video — event photography, product shots, headshots — and right now that material sits outside the process entirely. I have to bring photos in manually in Premiere afterwards, which breaks the flow of an otherwise automated assembly. Since Selects can already analyse video visually, extending that to images feels like a natural next step, and it would significantly help for event recaps and podcast episodes where we cover gaps with stills. That's the one gap I keep hitting. Review collected by and hosted on G2.com.