Users and Use Cases¶
Last verified: 2026-07-26 against dev at f0388abb.
This page describes product-facing users. Exact API auth and organization guardrails are covered in Tenancy and permissions.
User Groups¶
| User group | What they need |
|---|---|
| Analyst | Upload document packages, review extracted data, correct fields, validate results, and export outputs. |
| Program/operator user | Track assessment status and understand what documents still need processing or review. |
| Schema author | Define the structured fields the product should extract for a document type. |
| Extractor developer | Build or modify extractor workers that satisfy the extractor protocol and deployment contract. |
| Consumer app developer | Integrate another app with SDP through backend/server-side calls. |
| Tenant/platform engineer | Enforce organization access, configure environments, and deploy extractor infrastructure. |
| Evaluator | Run Phoenix/Arize experiments to measure extraction and checklist quality. |
Core Use Cases¶
Process an Assessment¶
An analyst creates an assessment, uploads files, lets the system process eligible documents, then reviews and validates extracted items.
Relevant docs:
Add a New Extraction Shape¶
A schema author or developer defines a schema, connects it to extractors through processor registration, and verifies the output through tests or experiments.
Relevant docs:
Add a New Extractor¶
An extractor developer implements an extractor app, connects the Service Bus queue and backend registry, and deploys matching infrastructure.
Relevant docs:
Integrate a Consumer App¶
A consumer app calls SDP from a server-side backend/BFF, forwards caller auth, and passes the consumer-owned LLM key only from server to server.
Relevant docs:
Measure Quality¶
An evaluator registers datasets, runs extraction/checklist experiments, and reviews scores, traces, latency, and output drift.
Relevant docs: