The newest stored answer for each of your industry's prompts, from each AI model at each location, with the full response text — and for every answer, whether it named your brand, where in the answer, and in what tone. This is how you find the prompts the AI answers without you, and read exactly what it says when it does.

Only your organization's own industry can be read here; any other id is a 404. Requires the GEO product.

Each prompt × model × location has exactly one newest answer, so there is no date window: since hides combinations whose newest answer is older than that day rather than returning older answers. Ask "which prompts does this model answer without naming me" with mentioned=false and models set to that one model — every other model's answer to the same prompt is a separate row, and you have not asked for it.

Answers where the model returned nothing are left out unless you pass include_empty=true. An answer with no text names no brands, so it would otherwise read as an answer that left you out.

The same brand is named in one answer as Rival Bank and in another as Rival; brands[].key is the same for both, so you can group on it. Paged, newest first. limit and offset are validated rather than adjusted: a limit above 100 is a 400, not a silently smaller page.

{
  "total": 47,
  "limit": 25,
  "answers": [
    {
      "prompt_text": "Which bank has the best mortgage rates?",
      "model": "chatgpt",
      "location": "US",
      "rank": 2,
      "sentiment": "positive",
      "sov_pct": 25,
      "brand_count": 7,
      "brands": [
        {
          "key": "rival bank",
          "name": "Rival Bank",
          "rank": 1,
          "mentions": 3,
          "sentiment": "positive"
        }
      ],
      "citation_count": 9,
      "source_count": 12,
      "citations": [
        {
          "domain": "nerdwallet.com"
        }
      ]
    }
  ]
}