Workflows and examples

By
The Quable Team

This page shows how the assistant chains the tools for common tasks, with prompts you can reuse. You write the request in plain language; the tool sequences are shown so you know what to expect and can steer the assistant when needed.


Start every session with pim-bootstrap

Get the pim-bootstrap prompt from the Quable server and follow its instructions.

The assistant calls, in order:

  1. get_user_info: your locales and permissions.
  2. list_locales: the active locales of the instance.
  3. list_document_types: document types and link types.
  4. list_completenesses: completeness rules.

From then on, it uses exact codes (fr_FR, reference_export_mkp_zal) instead of labels (French, Zalando). This avoids most errors.


Find products

By completeness:

Find all references with a Zalando completeness below 80% in French.

list_completenesses → search_products(document_type_id, completeness_id, completeness_locale: "fr_FR", completeness_value: "80", completeness_operator: "lt")

By attribute value:

List the jackets whose color is red or blue and that have no description in German.

list_attributes(search: "color") → list_predefined-values(attribute_id) to get the option codes → search_products(attribute_filters: {"product_color": ["red", "blue"], ...})

By category:

How many products are in the "Outdoor > Jackets" category?

search_classification(query: "jackets", catalog_id: "PIM") → pick the right path → search_products(classification_id, limit: 1) and read total.

With a link to the PIM:

Give me the 5 most recently modified products, with a link to each product page.

search_products(sort_by: "dateModified", limit: 5) → get_pim_url(entity_type: "product", legacy_id) for each item.


Enrich product data

Read the English description of product REF-001, translate it into German and Spanish,
show me the result, then save it once I approve.
  1. get_product(product_id: "REF-001", views: ["attributes"], attributes: "product_description")
  2. The assistant writes the translations and shows them to you.
  3. After your approval: update_product(document_id: "REF-001", attributes: '{"attributes": {"product_description": {"de_DE": "...", "es_ES": "..."}}}')

Only the locales sent are changed. If a value is not stored by the PIM (for example a character limit), the tool returns an error.


Create a product and its variants

Create a product "JKT-2027" of type reference, named "Winter jacket", in the category
Outdoor > Jackets, with the attribute set for jackets. Then create its variants
JKT-2027-S, JKT-2027-M and JKT-2027-L.
  1. list_document_types → check that the type accepts variants (has_variant).
  2. search_classification(query: "jackets") and search_attribute-sets(document_type_id).
  3. create_product(...): the result lists missing_required attributes and values not_persisted.
  4. create_variant(product_code: "JKT-2027", variant_code: "JKT-2027-S", ...) for each size.
  5. search_variants(product_code: "JKT-2027") to check.

  1. search_asset(orphans_only: true, created_after: "2026-11-16")
  2. For each asset, search_products(search: ["tshirt_red", "4568"]) with tokens from the filename. The result gives scored matches and no_match.
  3. The assistant shows you the proposed pairs.
  4. After your approval: add_asset_to_product(asset_code, document_code) for each pair.

Tag assets from their content

Look at the images in the DAM folder "Summer 2027", and tag each one with its season
in the attribute qua_asset_multicombo. Create the options you need.
  1. search_asset(classification: "...")
  2. get_asset_file(asset_id): the assistant sees the thumbnail.
  3. list_predefined-values(attribute_id: "qua_asset_multicombo") to reuse existing options.
  4. update_predefined-value(attribute_id, code: "season_summer", labels: '{"en_GB": "Summer", "fr_FR": "Été"}') if an option is missing.
  5. update_asset(asset_code, attributes: '{"qua_asset_multicombo": [{"id": "season_summer"}]}')

Read documents stored in the DAM

Read the supplier price list "prices_2027.xlsx" in the DAM and update the purchase price
of each listed product.

search_asset(original_filename: "prices_2027.xlsx") → get_asset_file(asset_id) returns the rows of each sheet → update_product for each product, after your approval.

For a PDF technical sheet, get_asset_file returns its text. Ask for mode: "document" when the layout matters (tables) or when the PDF is a scan.


list_document_types (to get the link type id) → add_link(origin: "JKT-2025", target: "JKT-2026", link_type: "replacement")

To remove it: remove_link(origin, target, link_type).


Export your catalog

Exports are asynchronous: the assistant starts the export, polls its status, then downloads the file.

Export all reference products to CSV, in French and English.
1. get_export_options()                         → allowed models (optional)
2. trigger_export(export_profile='{"model":"document.reference","extension":"csv","locales":["fr_FR","en_GB"]}')
                                                → process_id: "abc-123"
3. get_export_status(process_id="abc-123")     → status: running   (poll every ~10 s)
4. get_export_status(process_id="abc-123")     → status: finished, files: [{id: "file-uuid"}]
5. get_export_file(file_id="file-uuid")        → the CSV file

To export only some columns, ask for them explicitly: the assistant then calls get_export_options(model: "document.reference") to get the column codes and adds mappings to the profile. To reuse a profile saved in the PIM:

Run the export profile "Marketplace feed" and send the file to sftp://partner.example.com/in.

list_export_profiles(search: "Marketplace") → trigger_export(export_profile_id, upload_location: "sftp://...")

Files above 250 KB cannot be returned in the conversation: use upload_location, reduce the scope, or download the file from the Quable interface. See Limits.


Change the data model

Add the attributes attr_lining and attr_waterproof to the attribute set "jackets".

search_attribute-sets(id: "jackets") → the assistant shows the current composition → update_attribute-set(code: "jackets", attributes: '[...current..., "attr_lining", "attr_waterproof"]') → search_attribute-sets to check.