Mindspase is a visual knowledge base that auto-organizes everything you save: articles, images, videos, music, quotes, PDFs. No folders, no manual tags. It also has natural-language + visual search, shared spaces called Collective Minds, and end-to-end encryption with no ads and no data selling.

🔥 Our Take: I have never met a tagging system I kept up with. I save things when I’m busy, then later I remember the vibe, not the filename. This is betting on that being normal, and honestly… fair. If it can reliably pull the right thing from a half-memory, it’s useful. If it can’t, it’s just a prettier graveyard.

CY started a thread asking how people actually scan Product Hunt when the leaderboard is packed. What makes you click and what makes you scroll past.

The answers are basically a messy human checklist. People want a tagline that tells them the outcome fast, not a puzzle. Name and logo matter more than you’d think. Social proof matters too, a few folks admitted they wait until something has a minimum number of upvotes or comments before giving it attention, or they look for familiar upvoters. And there’s a growing anti-filter where if the name or tagline screams AI, some people just skip on principle because it all blurs together. 

Cardboard is a browser video editor where you tell it what you want and it builds the first cut for you. It can understand what’s in your clips, search footage by content, and supports live collaboration so someone else can jump in and review with you.

🔥 Our Take: I want this for the annoying parts of editing, cutting dead air, pulling selects, tightening pacing, getting to something watchable fast. The only thing that matters is whether the result stays easy to tweak after the magic runs. 

TADA is Hume’s open-source text-to-speech model built around 1:1 text-acoustic alignment, so the audio stays locked to the words. They claim about 5x faster generation, virtually zero word hallucinations or skipped content, and a footprint small enough for on-device use. 

🔥 Our Take: I care less about new voice demos and more about the boring failure mode where TTS casually drops a word and keeps going like nothing happened. The 1 to 1 alignment idea is nerdy, but it goes straight at that problem and also makes it faster. Open-sourcing it is a strong move.