The annoying part is that the signal you care about isn't in the metadata or the captions. It's inside the videos: what's said out loud, what's on screen, who they're actually talking to. So the easy-to-scrape stuff is useless and the useful stuff is expensive to look at.
So we built a pipeline that ingests TikTok and Instagram video and runs vision plus transcription over it, so you're querying the actual content instead of the classic metadata. This free (and no sign up) tool is a thin front-end over that same API.
How it works: paste an account, describe in plain language what you actually care about (not a fixed checklist), and it returns a score plus the specific moments behind it. Export to PDF if you want to hand it to someone. No login, no signup, no sales call.
The larger idea I'm chasing: most of the "marketing" work around content is really just watching, tagging, and filtering at a scale humans can't stand to do. That's exactly the kind of thing you can hand to a machine and get most of your time back. This tool is one slice of that.
Caveats, because you'll find them anyway: public accounts only (and we may not have some accounts), and it samples the last 3 months of videos rather than the full backlog.
Would love feedback on where the scoring feels wrong or misses context. And I'm curious how others here handle the content/creator side of growth without it eating your week.
slater•12h ago
yurimhln•12h ago