Collect Feedback for Presets

Send each user rating to POST /v1/feedback with the request ID and the preset that answered, then compare presets by how often their answers were rated well, not only by latency and cost.

Preset Feedback is in Alpha, and the API and dashboard may still change. Use the Feedback button on the Preset Feedback page to tell us what would make it more useful.

Record a rating

You need an API key with the feedback:write capability, the request ID Phaseo returned, and the UUID of the preset. The preset ID is what puts the event into preset comparisons and cohort analysis.

cURL

curl https://api.phaseo.app/v1/feedback \
  -X POST \
  -H "Authorization: Bearer $PHASEO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "requestId": "gen_01K...",
    "presetId": "8f54b2d9-9ba3-4ced-8a7f-9bb70a44dbea",
    "rating": "thumbs_up",
    "score": 1,
    "comment": "The answer resolved the customer question.",
    "metadataDimensions": {
      "plan": "pro",
      "surface": "support_chat"
    }
  }'

A feedback event must target at least one request, session, preset, or test run. score is optional and must be between 0 and 1. Put low-cardinality values such as plan, deployment, locale, or application surface in metadataDimensions, so you can filter by them. Put arbitrary or high-cardinality context in metadata instead.

Choose a rating

Signal	`rating`
A simple user reaction	`thumbs_up`, `thumbs_down`
An evaluated answer	`correct`, `partly_correct`, `incorrect`
An operational problem	`bad_format`, `too_slow`, `too_expensive`
A safety outcome	`unsafe`, `refused_incorrectly`
Anything else	`not_helpful`, `other`

Prefer an explicit rating to a numeric score alone: preset comparisons are built on ratings, and scores are shown as supporting detail.

Compare presets

Open Settings → Presets → Feedback to:

compare positive, partial, and negative rates
select a baseline preset
filter by date, rating, and metadata dimensions
inspect the request, session, comments, and tags behind an event

The dashboard analyzes at most the 10,000 most recent matching feedback rows in the selected window.

Read feedback through the API

Reading needs feedback:read. Results are limited to the authenticated workspace and the presets the caller can see. List recent events:

curl "https://api.phaseo.app/v1/feedback?preset_id=PRESET_UUID&limit=100" \
  -H "Authorization: Bearer $PHASEO_API_KEY"

Get a summary computed on the server:

curl "https://api.phaseo.app/v1/feedback/summary?group_by=preset&since=2026-08-01T00:00:00Z" \
  -H "Authorization: Bearer $PHASEO_API_KEY"

APIs for test runs and custom outcome events are available, but their dashboard views are still changing. To roll a preset change out gradually, see Roll out presets and debug routing.

Last modified on November 5, 2026

Overview

Models & Routing

Features

Integrations

Workspace & Access

Reliability & Operations

Platform & Research

Migration Guides

Community

Collect Feedback for Presets

Record a rating

Choose a rating

Compare presets

Read feedback through the API

​Record a rating

​Choose a rating

​Compare presets

​Read feedback through the API

Record a rating

Choose a rating

Compare presets

Read feedback through the API