Application model and pipeline

This page maps Klaaro's core entities and the processing pipeline that runs on every upload. Use the application model to see how teams, datasets, documents, and records relate in the dashboard and /api/v1 — then follow the Document Engine diagram for the ordered steps from ingest through extraction.

Application model

A team is the top-level tenant: members, roles, and API keys. Each dataset belongs to one team and holds pipeline configuration (classes and per-type schemas), ingested documents, and the records extracted from them. Classes label document types; schemas define the fields extraction writes into records.

Teammembers · API keys

owns

Datasetpipeline config

configures

Classdocument type

uses

Schemaextraction fields

1contains*

1conforms to*

Documentuploaded file

extracts

Recordstructured data

Teammembers · API keys

1owns*

Datasetpipeline config

1contains*

Documentuploaded file

1extracts*

Recorddata · schema

1configures*

Classdocument type

1uses1

Schemafields

Document Engine

Upload triggers the pipeline without extra API calls. Each document moves through parse and OCR, then classification and extraction against the dataset's classes and schemas. Results surface in Records (grid, insights, review) and as structured records via the API. Re-run from any step is possible.

Upload

Ingest documents of any kind from DMS, via SDK or API or web upload.

VARIOUS FILE TYPESHIGHLY PARALLEL PROCESSING

Parsing

Understands layouts, segments splitting, routing segments to best model.

Text extractionHandwritingSegmentation

Classify

Assign to one or more classes.

Create schema

Auto generate schema for new classes.

Extraction

Extract records specified in class schema.

LARGE LANGUAGE MODELSVISION MODELS

Records

Explore extracted records and apply human review. Annotate, review and approve extracted data.

ExploreVerifyApproveExport