Enterprise Document Processing
Reference solution for processing one inbound business document: parsing it, understanding it, classifying it, extracting structured information and producing a structured result with an explanation.
Unit event
Business Document
BUSINESS_DOCUMENTOne completed execution of the document-processing workflow for one inbound business document.
Design variant
A reasoning document-understanding pass, classification, two extraction passes, enterprise-knowledge grounding with reranking, and a grounded final generation. The same business unit event as Standard Processing, implemented differently.
Scenario inputs
Every value this design resolves, and who owns it. Change an editable value and the figures recalculate. The pricing engine answers, not this page: nothing here multiplies anything, and until it answers the figures beside you are the ones you were already looking at, labelled as such.
63 fields
Ingest & Understand Document12 fields
Pages in the inbound document. This is the metered quantity for document-page pricing.
Estimated tokens of text recovered from the document, published downstream so later activities do not restate it.
Input tokens eligible for cached-input treatment.
Estimated generated summary tokens, published downstream as an input size.
Estimated tokens of parsed document text presented to the model.
Estimated system/policy prompt tokens.
Classify Document7 fields
Input tokens eligible for cached-input treatment.
Estimated tokens of the summary presented for classification.
Estimated generated classification tokens.
Estimated system/policy prompt tokens, including the class taxonomy.
Extract Structured Information12 fields
Input tokens eligible for cached-input treatment.
Estimated tokens of the class and its field schema.
Estimated tokens of parsed document text presented for extraction.
Estimated generated extraction tokens.
How many fields the class asks for. A stated assumption of the scenario, recorded here rather than inside a pricing request definition.
Estimated system/policy prompt tokens.
Estimated tokens of the normalized field set, published downstream as an input size.
Retrieve Supporting Knowledge23 fields
Estimated tokens of the retrieval query, published downstream.
Embedding input tokens eligible for cached treatment.
Tokens submitted to the embedding model.
Number of candidate documents/chunks returned.
Estimated total tokens across candidate documents/chunks presented to reranking.
Number of candidate documents/chunks scored for the query.
Estimated total candidate-document tokens supplied to reranking.
Number of rerank queries.
Output tokens if the selected rerank commercial path meters them; zero in the pilot default.
Estimated tokens in the rerank query.
Estimated token size of assembled grounded context.
Produce Final Result9 fields
Input tokens eligible for cached-input treatment.
Estimated tokens of the document summary presented as context.
Estimated generated result and explanation tokens.
Estimated tokens of the normalized field set presented to the model.
Estimated tokens of grounded context presented to the model. Zero when the variant does not ground, which adds nothing to the summed input.
Estimated system/policy prompt tokens, including the result schema.
Cost by operation
Where the direct unit cost comes from. Expand an operation for the activities inside it and how each one resolved.
Ingest & Understand DocumentComplete2 activities, 2 priced0.16246 USD76% of direct total
Parse the inbound document into text and structure, then form a compact understanding of it.
Ingest and Parse Document
PricedRead an inbound business document and produce its text and layout structure. Commercially this is a document service metered by page, not a language model.
A pricing request was made and returned an amount.
Understand Document
PricedForm a compact understanding of the parsed document, sufficient for classification and for explaining the final result.
A pricing request was made and returned an amount.
Classify DocumentComplete1 activity, 1 priced0.002128 USD1% of direct total
Decide what kind of business document this is, which is what tells extraction what to look for.
Classify Document
PricedDecide the business document class, which determines what structured extraction looks for.
A pricing request was made and returned an amount.
Extract Structured InformationComplete2 activities, 1 priced0.0196 USD9% of direct total
Pull the fields the document class calls for, twice, then normalize them into the canonical shape.
Extract Structured Fields
Priced×2Extract the fields the document class calls for, from the parsed document text.
A pricing request was made and returned an amount.
Normalize Extracted Result
No pricing requestCoerce extracted field values into the canonical shape, types and units. Application logic, not a priced resource. Deliberately not ad-normalize-operational-result, which normalizes an operational system's reply and shares neither its inputs nor its outputs.
This activity does not price. No request was made, so there is no amount — not an amount of zero.
Retrieve Supporting KnowledgeComplete5 activities, 2 priced0.00082 USD0% of direct total
Ground the result against enterprise knowledge: build a query, embed it, search, rerank and assemble.
Prepare Document Retrieval Query
No pricing requestBuild an enterprise-knowledge retrieval query from the document summary. Deliberately not ad-prepare-retrieval-query, which builds one from a normalized customer request: same shape, different semantic input, and decisions §4 puts semantic correctness above the reuse ratio.
This activity does not price. No request was made, so there is no amount — not an amount of zero.
Embed Query
PricedGenerate a vector embedding for the retrieval query.
A pricing request was made and returned an amount.
Search Enterprise Knowledge
No pricing requestSearch enterprise knowledge using the query representation.
This activity does not price. No request was made, so there is no amount — not an amount of zero.
Rerank Candidates
PricedRerank retrieved candidates for relevance to the support query.
A pricing request was made and returned an amount.
Assemble Grounded Context
No pricing requestAssemble selected evidence into a bounded context for final response generation.
This activity does not price. No request was made, so there is no amount — not an amount of zero.
Produce Final ResultComplete1 activity, 1 priced0.02828 USD13% of direct total
Produce the structured result and a grounded explanation of how it was arrived at.
Generate Structured Result
PricedProduce the final structured result and a short explanation of how it was arrived at.
A pricing request was made and returned an amount.
Direct unit total
0.213288 USDBars show each operation’s share of the stated direct unit total. An operation with no priced activity shows no amount, which is not an amount of zero.
Assumptions, issues and reproducibility
Activity states not exercised by this configuration
The read model reports these so their absence reads as absence rather than as coverage.
Issues
The runtime reported none for this configuration.
Technical detail and reproducibility
- Result source
- REFERENCE_PREFETCH
- Scenario configuration
- scenario:enterprise-document:high-accuracy:2026-09-15
- Pricing effective at
- 2026-09-15T00:00:00Z
- Definition fingerprint
- d82b403b7d3a8653dccdac867e5dcb20f5e25ef57c9fe789887b7ca6ab811d0b
- Catalogue fingerprint
- fb4a33fb254bba125309001aadb10f470d9bf9f219404324ba86b2ad91c206f1
- Pricing engine version
- 0.7.0
- Pricing orchestration contract version
- v4
- Pricing rule registry version
- daf63fc55107b886370c9744e36b7276da8f235e6e5f8eb56a1d2214151568f5
- Scenario contract version
- v1
- Variant definition fingerprint
- d82b403b7d3a8653dccdac867e5dcb20f5e25ef57c9fe789887b7ca6ab811d0b