Skip to content

TRIDOT / DEVELOPERS / PLAYGROUND

Put this in, get this out.

Each scenario below was run on the real engine and its output captured. Pick one to see the input, the output and the same request as code. Sign in to change the input and run it yourself in the portal playground; playground runs are free and never billed.

Real outputs, scenario by scenario

Real engine output, captured 2026-10-04. Every input is a made-up scenario, not real user data.

Free · never billed

Scenario

Everyday small talk: every category should stay at none, with no action.

What to look forseverity none, action none and no score bars.

Sign in to run your own input

Real engine output, captured Oct 1, 2026

  1. RequestAssess the user turnPOST /v1/conversation-analytics/assess

    Input

    Just got off work, grabbing tacos with a friend tonight! What's your favorite food?

    {
      "event": {
        "eventId": "evt_6nga426j_0",
        "conversationId": "conv_6nga426j",
        "userIdHash": "u_6nga426j",
        "turnIndex": 0,
        "speakerType": "user",
        "text": "Just got off work, grabbing tacos with a friend tonight! What's your favorite food?",
        "timestamp": "2026-10-01T00:37:22.280Z",
        "characterId": "char_haru"
      }
    }

    Output

    Captured · Oct 1, 2026
    Severity noneAction none

    All 9 categories at 0 · none: self_harm, minor_safety, sexual, violence, harassment, pii, dependency, jailbreak, illegal

    Conversation peak
    none
    Conversation now (EMA)
    none
    • Engine 830 ms
    • Stages rules → llm
    • Model gpt-4.1-mini