AI Enrichment

AI Enrichment automatically generates and improves product content — titles, descriptions, tags, and SEO keywords — using large language models. You choose the AI provider and model, and Pixee PIM handles batching, cost tracking, and result storage.

GET/ai-providers/

List AI providers

Returns all configured AI providers and their status.

Request

GET

/ai-providers/

curl https://api.pixeepim.com/api/v1/ai-providers/ \
  -H "Authorization: Bearer {api_key}"

Response

{
  "providers": [
    {
      "id": "claude",
      "name": "Anthropic Claude",
      "status": "active",
      "models": ["claude-opus-4-7", "claude-sonnet-4-6"],
      "cost_per_1k_tokens": 0.015,
      "latency_ms": 300
    },
    {
      "id": "openai",
      "name": "OpenAI",
      "status": "active",
      "models": ["gpt-4o", "gpt-4o-mini"],
      "cost_per_1k_tokens": 0.03,
      "latency_ms": 250
    }
  ]
}

POST/enrichment/batch

Create an enrichment batch job

Creates and queues a batch enrichment job for a set of products.

Body parameters

Name
name
Type
string
Description
Descriptive name for the job.
Name
description
Type
string
Description
Optional longer description of the job's purpose.
Name
filters
Type
object
Description
Filter object to select which products to enrich (e.g. {"is_active": true, "supplier_id": "..."}). Enriches all products if omitted.
Name
enrichmentTypes
Type
array
Description
List of content types to generate. Accepted values: description, title, tags, seo_keywords, long_description.
Name
scheduled_at
Type
string
Description
ISO 8601 datetime to schedule the job for later execution. Starts immediately if omitted.

Request

POST

/enrichment/batch

curl -X POST https://api.pixeepim.com/api/v1/enrichment/batch \
  -H "Authorization: Bearer {api_key}" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "SEO Descriptions – Active Products",
    "description": "Generate SEO-optimised French descriptions for all active products",
    "filters": {"is_active": true},
    "enrichmentTypes": ["description", "seo_keywords"]
  }'

Response

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "name": "SEO Descriptions – Active Products",
  "status": "pending",
  "total_products": 842,
  "estimated_cost_usd": 3.80,
  "created_at": "2026-03-15T09:00:00Z"
}

GET/enrichment/batch/{job_id}

Get batch job status

Returns the current status and progress of a batch enrichment job.

Path parameters

Name
job_id
Type
string
Description
The batch job UUID.

Query parameters

Name
status
Type
string
Description
Filter jobs by status when listing: pending, running, completed, failed, paused, cancelled.
Name
limit
Type
integer
Description
Maximum number of jobs to return when listing (default: 20).
Name
offset
Type
integer
Description
Pagination offset when listing (default: 0).

To control a running job, use the dedicated action endpoints: POST /enrichment/batch/{job_id}/start, /pause, /resume, or /cancel.

Request

GET

/enrichment/batch/{job_id}

curl https://api.pixeepim.com/api/v1/enrichment/batch/550e8400-e29b-41d4-a716-446655440000 \
  -H "Authorization: Bearer {api_key}"

Response

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "name": "SEO Descriptions – Active Products",
  "status": "completed",
  "total_products": 842,
  "processed": 842,
  "failed": 3,
  "total_cost_usd": 3.74,
  "created_at": "2026-03-15T09:00:00Z",
  "completed_at": "2026-03-15T11:30:00Z"
}

POST/enrichment/products/{product_id}

Enrich a single product

Triggers AI enrichment for a single product and returns the results. Useful for on-demand enrichment from your own workflows.

Path parameters

Name
product_id
Type
string
Description
The product UUID.

Query parameters

Name
types
Type
array
Description
Comma-separated list of enrichment types to run: description, title, tags, seo_keywords, long_description. Runs all configured types if omitted.
Name
auto_apply
Type
boolean
Description
If true, applies the enriched fields directly to the product without a manual apply step (default: false).

Use GET /enrichment/products/{product_id} to retrieve stored enrichment results, GET /enrichment/products/{product_id}/progress to poll progress, and POST /enrichment/products/{product_id}/apply?fields=description,title to apply specific fields selectively.

Request

POST

/enrichment/products/{product_id}

curl -X POST "https://api.pixeepim.com/api/v1/enrichment/products/550e8400-e29b-41d4-a716-000000000001?types=description,seo_keywords&auto_apply=false" \
  -H "Authorization: Bearer {api_key}"

Response

{
  "product_id": "550e8400-e29b-41d4-a716-000000000001",
  "status": "queued",
  "types": ["description", "seo_keywords"],
  "auto_apply": false,
  "created_at": "2026-03-15T10:00:00Z"
}

POST/enrichment/products/{product_id}/from-documents

Document-based enrichment (free)

Enriches a product using only your previously indexed documents — datasheets, certificates, manuals — via Qdrant vector search (RAG). No LLM token cost is incurred for the retrieval step itself.

This endpoint is ideal when you have rich supplier documentation already indexed via Document Intelligence and want to populate product fields at zero AI cost.

Path parameters

Name
product_id
Type
string
Description
The product UUID.

Query parameters

Name
auto_apply
Type
boolean
Description
If true, applies the enriched fields directly to the product (default: false).

Requires documents previously indexed via Document Intelligence (/document-intelligence). The pipeline performs a Qdrant semantic search against the pm_products collection (512-dim, Cosine similarity) and synthesises the retrieved chunks into structured product fields. If fewer than 3 relevant chunks are found, the result may be incomplete — consider running a standard web enrichment job instead.

Request

POST

/enrichment/products/{product_id}/from-documents

curl -X POST "https://api.pixeepim.com/api/v1/enrichment/products/550e8400-e29b-41d4-a716-000000000001/from-documents?auto_apply=true" \
  -H "Authorization: Bearer {api_key}"

Response

{
  "product_id": "550e8400-e29b-41d4-a716-000000000001",
  "enrichment_type": "document_rag",
  "rag_chunks_found": 5,
  "has_web": false,
  "auto_apply": true,
  "phases": {
    "collect": "completed",
    "synthesize": "completed"
  },
  "result": {
    "seo_description": "Produit certifié CE conforme à la directive RoHS, conçu pour une utilisation industrielle.",
    "technical_summary": "Tension 230V, puissance 1500W, IP44, dimensions 200×150×80 mm."
  },
  "created_at": "2026-03-15T10:05:00Z"
}

GET/ai-services/recommendations/{use_case}

Provider recommendations

Returns ranked AI provider recommendations for a specific enrichment use case, based on quality, cost, and speed benchmarks.

Path parameters

Name
use_case
Type
string
Description
One of: description_longue_seo, title_optimization, tags_generation, seo_keywords.

Query parameters

Name
max_cost
Type
number
Description
Maximum acceptable cost per 1k tokens (USD).
Name
min_quality
Type
number
Description
Minimum quality score 0–10.

Request

GET

/ai-services/recommendations/{use_case}

curl "https://api.pixeepim.com/api/v1/ai-services/recommendations/description_longue_seo?max_cost=0.02" \
  -H "Authorization: Bearer {api_key}"

Response

{
  "use_case": "description_longue_seo",
  "recommendations": [
    {
      "provider_name": "Claude Sonnet",
      "cost_per_1k_tokens": 0.015,
      "quality_score": 9.2,
      "speed_ms": 2500,
      "recommended": true
    },
    {
      "provider_name": "GPT-4o Mini",
      "cost_per_1k_tokens": 0.006,
      "quality_score": 7.8,
      "speed_ms": 800,
      "recommended": false
    }
  ]
}