---
title: "Refresh Embeddings Endpoint"
method: POST
path: "/api/v1/remy/cron/embeddings/refresh"
tags: ["remy"]
---

# Refresh Embeddings Endpoint

`POST /api/v1/remy/cron/embeddings/refresh`

## Request

- Base URL: `https://api.brandbooster.ai` (the document declares no server; this is the origin it was published from)
- URL: `https://api.brandbooster.ai/api/v1/remy/cron/embeddings/refresh`
- Auth: none declared

Kick off today's AI Search embedding refresh and return at once. Runs after the scrapes.

Two passes. Creators with no embedding at all are cleared first — until a creator has one they
are invisible to semantic narrowing, and the scrapers add roughly 5,000 a day. Then a capped
pass over creators whose profile row has moved since their embedding was written, so an
enriched bio eventually gets a vector that matches it.

Safe to run repeatedly and safe to miss: candidates are computed from the data rather than from
a schedule, so a second run the same day is close to a no-op and a missed night is picked up by
the next one. Queued — the initial backfill of 24,673 creators took 24 minutes.

## Response `200`

Successful Response

- StartEmbeddingRefreshResponseDTO
  - `job_id` string, required
  - `status` string, required

## Changes

> 37 revisions in range; 1 not diffed.

- **2026-09-16** `dda5cf04b73d`: 1 info
  - endpoint added

[Operation history](https://skmtc.dev/brandbooster/apis/brandbooster-api/changes/api/v1/remy/cron/embeddings/refresh/post.md)

---

[API](https://skmtc.dev/brandbooster/apis/brandbooster-api.md) - [All operations](https://skmtc.dev/brandbooster/apis/brandbooster-api/llms.txt) - [OpenAPI document](https://skmtc.dev/brandbooster/apis/brandbooster-api/revisions/02d28bde684c?raw)