Returns a paginated list of CSV batch classification jobs, sorted by creation date descending.

Maximum number of records to return (page size)

$top

Number of records to skip (offset); use multiples of $top

$skip

Unique correlation ID for distributed tracing. If not provided, the server generates one.

X-Correlation-Id

Client identifier in the form '{app}; {version}; {adapter}; {machine}'. Used for telemetry and troubleshooting.

X-Avalara-Client

API version. Required for all Avalara public APIs.

avalara-version

Forbidden — the caller is authenticated but lacks permission for this resource.

Too Many Requests — client rate limit exceeded. Retry after the interval indicated by the Retry-After header.

Internal Server Error — an unexpected error occurred.

List all classification batches for the current tenant.

Parses the uploaded CSV and queues product and worksheet creation asynchronously. Required columns: productNumber, description, countryOfOrigin, countryOfDestination, poNumber. Returns `202 Accepted` with an `Operation-Location` header pointing at the batch status URL. Per-item results for both successful and unsuccessful rows are available by polling that URL; each result correlates with the original CSV row by its `rowIndex` (0-based position in the uploaded file), or by `productNumber`.

Submit a batch of products for classification via CSV upload.

Returns processing status, aggregate counts, and per-item results for both successful and unsuccessful items. Each result can be correlated with the original CSV row by `rowIndex` (0-based position in the file) or by `productNumber`.

batchId

Get the status of a classification batch.

Enhances the supplied description into customs-grade wording with quality scores before and after. An HS code is classified and returned only when the request explicitly selects a classification provider (e.g. HSAC); when provider is omitted no HS code is returned. Returns 204 when no useful suggestion is produced; the reason is carried in the X-Description-Reason header. An unsupported classification provider returns 501.

Enhanced description with quality scores; an HS code is included only when a provider was selected and classification succeeded

Suppressed; see X-Description-Reason header

Requested classification provider not supported

Enhance a product description and classify its HS code.

Returns a deterministic quality score (0-100) and per-signal breakdown for the supplied description. An empty description scores zero.

Uploads an image, extracts product features using vision AI, and returns HS code suggestions. The country code defaults to the worksheet's destination country if not provided.

Country of destination (ISO 2-letter). Defaults to worksheet country.

countryCode

Image classification is disabled by configuration

Image classification service temporarily unavailable

Classify a product from an image within a worksheet context.

Processes each image through the vision AI pipeline. The response array always contains one entry for every uploaded image, returned in the same order as the files were sent, so callers can correlate each result with its image by position (index) in the array.

Country of destination (ISO 2-letter). Required.

All images processed (individual results may vary)

No images provided or missing country code

Classify multiple product images using AI vision (standalone).

Uploads an image, extracts product features using vision AI, and returns HS code suggestions.

Country of destination (ISO 2-letter). Required for standalone.

Invalid image, missing country code, or unsupported format

Classify a product from an image (standalone, no worksheet).

Registers an uploaded file, linking it to a worksheet and optionally to a sub-part or a customs ruling. Sub-part and ruling associations are mutually exclusive.

Returns the identifier of the registered file.

Returns an error if the file registration request fails validation, such as when both a sub-part and a ruling are specified.

Returns an error if the request lacks a valid authentication token.

Downloads a previously uploaded file from S3 by its file ID.

fileId

Returns the requested file as a binary download.

Returns an error if the file is not found.

Deletes a file from S3 storage and removes its database record.

Returns a single classification worksheet by its unique identifier.

Returns an error if the worksheet is not found.

Updates worksheet fields such as sub-parts, classifications, GRI conditions, and PGA requirements.

Marks the worksheet as deleted. The record is retained with deletedAt and deletedBy set and is excluded from default list queries. Idempotent: deleting an already-deleted worksheet returns 204 without changing the record.

Worksheet deleted successfully (or already deleted)

Changes the current workflow step of the worksheet to the specified target step.

Returns the worksheet with the updated workflow step.

Returns the active match threshold and related options. Falls back to global defaults when no tenant override has been saved.

Returns the tenant's item-match settings.

Upserts the tenant's match threshold and related options. Null fields are left unchanged. The threshold must be greater than 0 and no more than 100.

Returns an error if the threshold or other provided values are invalid.

Update item-match settings for the tenant.

Returns a paginated list of classification worksheets ordered by ID descending.

Returns a paginated list of worksheets for the tenant.

Creates a new classification worksheet linked to the specified product ID.

Reverts a COMPLETED worksheet back to DRAFT and its linked product to IN_REVIEW. Decrements the PO's classifiedCount if applicable and writes the line status back to IN_REVIEW.

Reopen a completed worksheet for reclassification.

Looks up PGA rules for each classified HTS code and stores worksheet review rows.

Returns the worksheet with refreshed PGA requirements.

Returns an error if the worksheet is not found for the tenant.

Refresh PGA requirements for a worksheet.

Recomputes Same/Similar candidates and/or CROSS ruling evidence for this worksheet. Use stage=matches for the fast item-master phase, stage=rulings for the slower CROSS evidence phase, or stage=all for both.

Specifies which refresh stage to run: matches for the item-master phase, rulings for the CROSS evidence phase, or all for both.

stage

Returns the worksheet with refreshed match analysis.

Records that a named milestone has been completed on the worksheet.

Returns the result of the milestone recording operation.

Calls HSAC COO prediction API with the worksheet's product data and persists the predicted country on the worksheet if not already set.

COO prediction is disabled by configuration

Predict country of origin for a worksheet product.

Breaks down the product into sub-parts using Bill of Materials data.

Returns the worksheet with sub-parts populated from the Bill of Materials.

Generates GRI basis, supporting logic, description, and ruling fields using AI.

Generates a legal justification for the HTS classification using AI.

Generates relevant customs rulings for the HTS classification using AI.

Marks the worksheet as completed and saves the classification result.

Triggers AI-based HS code prediction for a specific sub-part within the worksheet.

Returns the AI-predicted classification for the sub-part.

Triggers AI-based HS code prediction for the main product description.

Returns the AI-predicted classification for the product.

Calculates duty rates and landed cost for a sub-part via the Quoting API.

Returns the calculated duty rates and landed cost for the sub-part.

Returns an error if the sub-part identifier is missing or the worksheet is in an invalid state for duty calculation.

Returns a paginated queue of products that have not yet been classified, ordered by priority. Supports optional search/filter parameters for narrowing results.

Full-text search on productNumber + description (case-insensitive)

search

priority

status

poNumber

Filter products with/without AI match candidates

hasMatches

Include only products from this creation channel

sourceChannel

Exclude products from this creation channel

excludeSourceChannel

Sort order as 'field,direction' (e.g. 'sortPriority,asc'). Supported fields: sortPriority, createdAt, priority. Defaults to priority ascending.

sort

Returns a paginated queue of products awaiting classification, ordered by priority.

Creates a new Product and its associated ClassificationWorksheet in a single call. Use this when no purchase order or intake request exists for the item.

Runs AI HS prediction over every pending product and persists the top suggestion on each worksheet, moving the products to in-review for human confirmation. The `results` array always contains one entry for every product processed — both successful items (`outcome=CLASSIFIED`) and unsuccessful ones (`outcome=SKIPPED` or `outcome=FAILED`); unsuccessful entries carry the reason in `error`. Callers can correlate each entry by its `rowIndex` (position in the results array) or by productId/productNumber. Aggregate counts (total, classified, skipped, failed) are also returned.

Bulk-classify pending products with Jarvis.

Fetches an item by its monolith ID, creates a local Product and ClassificationWorksheet, and links them to the original item for write-back on completion.

itemId

Create a worksheet from a monolith Item Master record.

Creates a Product and ClassificationWorksheet with sub-part materials in one call. Use classificationType to indicate BOM, KIT, or SPARE_PARTS; defaults to BOM.

Best-effort batch refresh of every purchase order already tracked for this tenant. Per-PO failures are skipped without aborting the batch. The `results` array reports the outcome for every purchase order processed — both successful (`status=SUCCESS`) and unsuccessful (`status=FAILED` or `status=SKIPPED`) items; every unsuccessful entry always carries a non-null `error` with the reason, while SUCCESS entries have a null `error`. Callers can correlate each entry by purchaseOrderId. Aggregate counts and the most recent sync timestamp are also returned.

Returns the aggregate sync counts and the most recent sync timestamp.

Returns an error if the monolith base URL is not configured.

Re-sync all known purchase orders for the tenant.

Fetches a purchase order by ID from the monolith and creates one ClassificationWorksheet per PO line, one Product per line, and a SyncedPurchaseOrder rollup header.

The unique identifier of the purchase order.

purchaseOrderId

Returns the created worksheets, products, and purchase order rollup summary.

Returns an error if the purchase order is not found.

Sync a purchase order and create worksheets.

Processes each PO-kit PDF through the IDP pipeline. The response array always contains one entry for every uploaded file, in the same order as the files were sent. Each entry's `status` distinguishes a successful extraction (`success`) from an unsuccessful one (`error`).

All PO-kit PDFs processed (individual results may vary)

Upload multiple PO-kit / retail-set PDFs for IDP extraction.

Uploads a PDF to the monolith IDP pipeline, routed to the PO-kit extractor (a retail set such as a sewing kit, distinct from a manufacturing Bill of Materials). Returns the kit's articles as structured data for the user to review before creating a KIT worksheet.

PO-kit PDF processed and articles extracted

Upload a PO-kit / retail-set PDF for IDP extraction.

Processes each BOM PDF through the IDP pipeline. The response array always contains one entry for every uploaded file, in the same order as the files were sent, so callers correlate each result by its index. Each entry's `status` distinguishes a successful extraction (`success`) from an unsuccessful one (`error`), and failed entries carry the reason in `error`.

All BOM PDFs processed (individual results may vary)

Upload multiple BOM PDFs for IDP extraction.

Uploads a PDF to the monolith IDP pipeline. Returns extracted materials/components as structured data for the user to review before creating sub-parts on a worksheet.

BOM PDF processed and materials extracted

Accepts a product description and destination country, calls Jarvis for HS classification, creates a Product and Worksheet from the results. Image URL and product URL are accepted for user reference context.

Classification completed, product and worksheet created

Classify a product from description and optional URLs (standalone).

Processes each PDF through the IDP pipeline and Jarvis classification. Returns aggregate success/failure counts plus a per-file `results` array; correlate each result with the uploaded files by index. Each file's `status` distinguishes success from failure, and failed files carry a reason in `error` without blocking successful files.

All PDFs processed (individual results may vary)

No files provided or missing country code

Upload multiple PDFs for IDP extraction and AI classification.

Uploads a PDF to the monolith IDP pipeline for text extraction, then classifies each extracted line item with Jarvis. Creates a Product and Worksheet per extracted item.

Upload a PDF for IDP extraction and AI classification.

Returns a printable summary of the classification decision and supporting data.

Returns the list of revisions recorded for the worksheet.

Returns all uploaded documents associated with the worksheet.

Returns the list of uploaded documents associated with the worksheet.

Returns a paginated list of purchase orders that have been synced, ordered by sync date descending.

Returns a paginated list of purchase orders synced for the tenant.

List synced purchase orders for the tenant.

Returns the list of products for a purchase order.

Returns the list of products for the purchase order.

Get products synced from a purchase order.

Returns all classification worksheets associated with the given purchase order (one per PO line).

Returns the classification worksheets associated with the purchase order.

Returns counts for each lifecycle stage (awaiting classification, in review, classified), how many pending products appeared since the supplied 'since' timestamp, the most recent purchase-order sync time, and new-parts counts segmented by purchase-order provenance and same/similar match quality.

ISO-8601 instant of the caller's last view; controls newSinceLastSeenCount

since

Dashboard summary of the classification queue.

Returns counts for the queue dashboard: total parts awaiting classification, parts with AI matches above the tenant threshold, parts with a purchase order, and the most recent purchase order sync timestamp for the tenant. The timestamp is empty if the tenant has never synced.

Returns the aggregated counts and last sync timestamp for the classification queue.

Aggregated counts for the classification queue.

Returns a paginated list of products that have completed classification. Supports optional search/filter parameters.

Returns a paginated list of products that have completed classification.

Returns milestone completions across all worksheets for the given product.

productId

Returns the milestone completions recorded across all worksheets for the product.

Returns the classification worksheet linked to an intake request, creating the worksheet from the intake request details when needed.

intakeRequestId

Resolve or create a worksheet for an intake request.

tradeHubWorksheet

A component part or article of an assembly (BOM/kit) extracted from a PDF

Per-file extraction outcome: 'success' when materials were extracted, 'error' when the file timed out or failed.

Target workflow step to move the worksheet to.

ISO-3166-1 alpha-2 destination country applied to every classified product. Falls back to each product's existing destination when omitted.

ISO-3166-1 alpha-2 destination country used for the classification. Falls back to the worksheet's destination when omitted.

Identifier of the product to create a classification worksheet for.

The sub-part and cost details to calculate duties for.

Portable automatic data processing machines, weighing not more than 10 kg

Portable digital automatic data processing machine

Classified HS/HTS code the content is generated for.

Tariff (HTS) description used as additional context for generation.

Item description used as additional context for generation.

The tenant's match threshold and related options to update.

Per-file outcomes, in the order the files were supplied; correlate with the request by index

Parent + materials extracted from a BOM/KIT PDF (BOM/KIT only)

Auto-detected document classification; always populated on success. SINGLE_ITEM -> worksheets populated; BOM/KIT -> assembly populated.

IDP document identifier for the uploaded PDF

Failure reason; null when status is 'success'

Per-file outcome. 'success' means the PDF was classified; 'error', 'timeout', and 'empty' are unsuccessful outcomes carrying a reason in 'error'.

Worksheets created from the extracted line items (SINGLE_ITEM only)

Failure or skip reason; null when outcome is CLASSIFIED

Business product number echoed for correlation

Zero-based position of this product in the processed collection, for correlating the result with the request by index

Worksheets created for this purchase order; zero unless status is SUCCESS

Failure reason; populated for FAILED/SKIPPED items, null for SUCCESS

Purchase order identifier echoed for correlation

Per-item outcome. Successful items are SUCCESS; unsuccessful items are FAILED (processing error) or SKIPPED (intentionally not processed), and always carry an error.

Worksheets updated for this purchase order; zero unless status is SUCCESS

TradeHub Classification Worksheet API