Back to IDA
Inventory · 02 of 04

Pipelines

Every data flow and ETL process that connects your systems — sources, targets, schedules, and frequencies — extracted from your integration docs.

Pipelines are the connective tissue. When they're catalogued, lineage falls out for free.

ETLStreamingAPI FeedsBatch Jobs
Book a demo

Pipelines

2 of 4 Inventories

91

records

54

Approved

28

Pending

9

Manual

Live counts from the agent pipeline

Data Flow

How records get here.

Records land in the inventory two ways — extracted from documents by an agent, or entered by your team. Both paths feed the same approval queue.

Agent

Pipeline Inventory Agent

AI-powered extraction.

The agent reads integration documents and extracts pipeline definitions, source-to-target mappings, and scheduling information automatically.

Processes these document types

Integration SpecsData Flow DiagramsETL DocumentationAPI Schemas

Automatic source-and-target detection across documents.

Human

Human in the loop

Manual entry and review.

Data engineers and stewards can add pipelines directly or review and validate agent-extracted records before they feed lineage.

Human actions

  • Add records manually
  • Edit existing data
  • Approve pending items
  • Remove duplicates

Every AI suggestion requires human approval.

Quality Indicator

Trust scores, by source.

Every record carries a trust score based on the maturity of the document it came from. Approved technical specifications produce high-trust records. Informal notes produce low-trust ones.

High Trust

From approved specs and signed-off requirements.

Medium Trust

From draft documents and work-in-progress specs.

Low Trust

From informal docs, emails, and early drafts.

Pilot distribution

High
55%
Medium
30%
Low
15%
Approval Workflow

From pending to production.

Records start as pending. They must be marked use before downstream tools like lineage can pick them up.

Pending

Awaiting review

Use

Ready for lineage

Lineage

Connected to pipelines

Interface Preview

The Pipelines table.

Track data flows with source-target mappings, frequencies, and full traceability — sortable, filterable, exportable.

Filter & SearchSort ColumnsInline EditingAdd RecordsDelete RowsExport CSV
6records3 approved3 pending
#
Trust
ActionPipeline IDPipeline NameSource SystemTarget SystemFrequencySource Documentation
1
High
UseCRM-DWH-001CRM Customer SyncCRM SystemData WarehouseDailyCRM Integration Spec v2.pdf
2
High
UseERP-DWH-001ERP Financial ExtractERP PlatformData WarehouseDailyERP Data Pipeline Requirements.docx
3
Medium
UseDWH-RPT-001Reporting AggregationData WarehouseReporting EngineHourlyReporting Pipeline Tech Spec.pdf
4
Medium
PendingKYC-CRM-001KYC Customer UpdateKYC SystemCRM SystemReal-timeKYC Integration Business Req.docx
5
Low
PendingATM-DWH-001ATM Transaction FeedATM SystemData WarehouseReal-timeATM Data Flow Specification.pdf
6
Low
PendingDWH-ML-001ML Feature PipelineData WarehouseML PipelineDailyManual entry via UI

This is a preview — the real interface supports full CRUD operations

The Complete Picture

Three more inventories.

Pipelines is one of four. Together they form your end-to-end landscape — from the source system down to the column in the database.

Where It Shows Up

Tools that depend on this inventory.

Pipelines are the edges in the lineage graph. The catalogue here drives every connection drawn there.

Get Started

Ready to map your data flows?

We'll point IDA at your integration docs in a 30-minute call and show you the pipeline catalog that comes out.

Book a demo