Platform

Eight stages.
One governed pipeline.

Every stage is agent-proposed and human-approved. Every decision is logged to the person who made it. The pipeline runs from raw extract to signed handoff without a single unreviewed transformation.

Ingest
Schema Detective
Normalise
Data Prep
Semantic Model
MDM
Target Mapping
Activate
Ingest

Pull from ERP extracts, CSV dumps, SaaS APIs, and database snapshots. PyAirbyte connectors for 500+ sources. Every record tagged with source and ingestion timestamp.

Input

CSV, JSON, SQL dumps, REST APIs, database connections

Output

Staged raw tables with _xyn_source + _xyn_ingested_at tags

The agent reads your source schemas and builds a structural map — data types, nullable columns, implicit FK relationships, naming conventions — before any transformation is written.

Input

Raw staged tables

Output

Schema report, structural fingerprint, FK graph

Detects type mismatches, composite keys, orphaned columns, and naming inconsistencies. Proposes a normalised structure. Engineer reviews and approves before any data moves.

Input

Schema report

Output

Normalisation proposal, approved DDL

LLM-assisted cleaning rules, format standardisation, PII detection and masking, null-handling strategies. Every rule proposed by the agent, approved by your team.

Input

Normalised staging tables

Output

Clean prep tables, cleaning audit log

The agent generates a semantic model (dbt metric layer) from the cleaned prep tables — dimension hierarchies, measure definitions, entity relationships. You approve the model before it deploys.

Input

Clean prep tables

Output

dbt semantic model, mart definitions

MDM →

Probabilistic entity resolution using Splink with configurable blocking rules and EM training. The agent surfaces duplicate candidate pairs. Stewards approve or reject merges. Every decision is logged.

Input

Clean prep tables

Output

Golden records, merge decision log, survivorship rules

Maps golden records to the target ERP schema — exact column match and LLM-assisted semantic mapping for non-obvious equivalences. Produces a signed mapping manifest for review.

Input

Golden records, target ERP schema

Output

Mapping manifest, value transformation rules

Packages the approved mapping manifest, golden records, and audit trail into a signed handoff ZIP ready for ERP load or mart deployment. The ERP consultant owns the final load step.

Input

Approved mapping manifest

Output

Handoff ZIP, migration_id_registry, deployment report

See the pipeline on
your data.