> ## Documentation Index > Fetch the complete documentation index at: https://geoptie.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # Update recommendations > Re-check this brand's recommendations against its latest AI answers, the same update as the dashboard's Check for updates button. Recommendations are built from the answers to your prompts, which come in once a day, so a brand can be updated once every 24 hours; sooner returns `429` with `Retry-After`. Recommendations also update by themselves when they are more than a week old and someone opens the Recommendations page. `mode: more` instead checks more pages for outreach, up to 3 times a day. Returns straight away with `poll_after` and a `poll` path; the update takes a few minutes. Poll the recommendations list and treat it as finished once `last_seen_at` is later than `poll_after`. ## OpenAPI ````yaml /openapi.json post /v1/brands/{id}/recommendations/generate openapi: 3.1.0 info: title: Geoptie API version: 1.0.0 summary: Read and manage your AI-search visibility data. description: >- Everything the Geoptie dashboard can do, over HTTP. **Authentication.** Every request needs an API key as `Authorization: Bearer `, and the workspace must have an active or trialing subscription. Keys are created in the dashboard and scoped to one workspace. **Pagination** is cursor-based. Pass `next_cursor` back as `cursor` and stop when `has_more` is false. The cursor is opaque; do not parse it. **Dates** are ISO 8601. `from` is inclusive, `to` is exclusive. Omitting both gives the last 30 days. **Limits** exist to stop runaway clients, not to meter you: there is no monthly request ceiling. Every response carries `X-RateLimit-*`, and endpoints that spend money carry `X-Endpoint-*` as well. A 429 names which limit was hit. **Errors** always carry a stable `error.code`. Match on that, never on the message text. A verb that an endpoint does not declare returns 405 `method_not_allowed`. contact: name: Geoptie support email: support@geoptie.com servers: - url: https://api.geoptie.com description: Production security: - bearerAuth: [] tags: - name: Brands description: The brands you track. A brand owns its prompts, engines and country. - name: Prompts description: >- The questions you track across engines, the answers they returned, and the brands named in those answers. Plan limits apply to how many you can track. - name: Visibility description: How often and how prominently a brand appears in engine answers. - name: Citations description: Which pages engines cited when answering your prompts. - name: Competitors description: >- Your share of voice against the other brands appearing in your prompts, the tracked set that share of voice is computed over, and per-competitor trends. - name: Topics description: Optional grouping for prompts within a brand. - name: Recommendations description: Generated actions for improving visibility, and their lifecycle. - name: Audit reports description: On-demand GEO analysis of a single URL. - name: Content generations description: >- New articles, written from the pages engines already cite for the prompts you track. Research a brief first, then write the article from it. - name: Content optimizations description: >- Pages you already have, scored for AI search with the specific changes that would raise the score. paths: /v1/brands/{id}/recommendations/generate: post: tags: - Recommendations summary: Update recommendations description: >- Re-check this brand's recommendations against its latest AI answers, the same update as the dashboard's Check for updates button. Recommendations are built from the answers to your prompts, which come in once a day, so a brand can be updated once every 24 hours; sooner returns `429` with `Retry-After`. Recommendations also update by themselves when they are more than a week old and someone opens the Recommendations page. `mode: more` instead checks more pages for outreach, up to 3 times a day. Returns straight away with `poll_after` and a `poll` path; the update takes a few minutes. Poll the recommendations list and treat it as finished once `last_seen_at` is later than `poll_after`. operationId: generateRecommendations parameters: - name: id in: path required: true description: Brand id. schema: type: string format: uuid example: b7c1e5a0-4f3d-42a8-9c6b-1e8d5a2f7043 requestBody: required: false content: application/json: schema: type: object properties: mode: type: string enum: - refresh - more default: refresh description: >- `refresh` re-checks everything. `more` checks more pages for outreach. responses: '202': description: Accepted content: application/json: schema: type: object required: - data properties: data: $ref: '#/components/schemas/RecommendationRunAccepted' '400': $ref: '#/components/responses/E400' '401': $ref: '#/components/responses/E401' '402': $ref: '#/components/responses/E402' '403': $ref: '#/components/responses/E403' '404': $ref: '#/components/responses/E404' '429': $ref: '#/components/responses/E429' '500': $ref: '#/components/responses/E500' components: schemas: RecommendationRunAccepted: type: object properties: brand_id: type: string format: uuid status: type: string enum: - generating mode: type: string enum: - refresh - more poll_after: type: string format: date-time description: >- The completion watermark. Poll `poll` and treat the run as finished once the recommendations' `last_seen_at` is later than this. poll: type: string description: Path to poll for the result. Error: type: object required: - error properties: error: type: object required: - code - message properties: code: type: string description: Stable machine code. Match on this, not the message. message: type: string responses: E400: description: >- Bad request. `missing_parameter`, `invalid_request`, `invalid_engine` or `invalid_date`. content: application/json: schema: $ref: '#/components/schemas/Error' E401: description: '`missing_api_key` or `invalid_api_key`.' content: application/json: schema: $ref: '#/components/schemas/Error' E402: description: >- `subscription_required`. The workspace has no active or trialing subscription. content: application/json: schema: $ref: '#/components/schemas/Error' E403: description: '`plan_not_eligible` or `forbidden`.' content: application/json: schema: $ref: '#/components/schemas/Error' E404: description: '`not_found`. Also returned for records belonging to another workspace.' content: application/json: schema: $ref: '#/components/schemas/Error' E429: description: >- `rate_limited`, `quota_exceeded`, `endpoint_rate_limited` or `concurrency_limited`. content: application/json: schema: $ref: '#/components/schemas/Error' E500: description: '`internal_error`. Quote the `X-Request-Id` when reporting it.' content: application/json: schema: $ref: '#/components/schemas/Error' securitySchemes: bearerAuth: type: http scheme: bearer description: 'Your API key, e.g. `Authorization: Bearer gp_live_...`.' ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.