Flaws in videos, ads and designs

Have people spot what an AI made wrong before it ships: clipped text, glitches, warped hands, captions that don't match.

How it works for this decision

Your agent sends

{
  "kind": "judgment",
  "format": "flaws",
  "question": "Does this video have a flaw?",
  "video_url": "https://example.com/ad.mp4",
  "media_seconds": 20,
  "reasons": "all"
}

A verified person sees

Task$0.05

Does this video have a flaw?

A 20 second product ad made by an AI video tool, sent as video_url.

You get back

{
  "answer": "flaw",
  "agreement": 0.8,
  "answers": 5,
  "votes": {
    "none": 1,
    "flaw": 4
  },
  "contested": false,
  "reasons": [
    {
      "answer": "flaw",
      "reason": "The caption at 0:12 says 30% but the voice says 20%."
    },
    {
      "answer": "flaw",
      "reason": "The woman's left hand has six fingers around 0:07."
    }
  ]
}

The result shown is an example. Try answering the task yourself.

Why a human check

Generated videos, ads and layouts often have flaws a person sees in a second and a model misses: a hand with six fingers, a caption a beat late, a headline cut off on mobile, text that's hard to read.

Several people checking the same piece and saying what's wrong is a fast, cheap last look. Ask for reasons on every answer and you get a list of what to fix.

Questions

Who checks it?

ID verified people from the general public, the same kind of people who will see your ad. Each one watches the clip through before answering.

How do I start?

Get your API key at plures.si. Start with your own team: the first 50 decisions each month are free. Add credit when you want ID verified people on the rest, live right away.

More use cases