Sentiment

Score how each AI model talks about a brand (positive / neutral / negative) and the verbatim phrases driving the tone.

Overview

Not just whether the models mention a brand, but how. Your agent gets a positive/neutral/negative split per model plus the phrases driving it, and an aggregate across all models queried.

Example

Ask your agent in plain English and it calls the tool for you:

# you say
How do the AI models talk about Attio as a CRM? Is the sentiment positive?

The tool returns structured JSON to your agent (you never see this directly):

{
  "brand": "attio",
  "perModel": [
    { "model": "chatgpt", "positive": 0.55, "phrases": ["modern, AI-native CRM"] }
  ],
  "aggregate": { "positive": 0.62, "negative": 0.08 },
  "billing": { "cost": { "usd": "0.36" } }
}

Your agent reads that JSON and answers you in plain English. That reply is written by your own AI; the JSON is just what it reasons over:

# your AI replies
Mostly positive: 62% positive, just 8% negative across the models. They reach for phrases like “modern, AI-native CRM” and “polished UI”. The main reservation is depth for large teams, so that’s the narrative to get ahead of.

Full input and output schema: POST /v1/sentiment. Pass an optional models list to narrow which models run (see per-model pricing).