The analytics stack for e-commerce, as one application.
ERP AI Layer replaces Fivetran, Snowflake, dbt and Looker for a mid-market e-commerce team: it ships the ingestion, the warehouse model, the metric definitions and the dashboards, rather than the tools to build them.
Its wedge is one number. Your ERP books revenue and product cost, but not outbound freight, payment and marketplace fees, or the eventual cost of returns — which on a DTC line routinely total 18–30% of net revenue. This decomposes all seven cost components per order line and reports true contribution margin.
Explore the seeded workspace
No password neededNorthwind Commerce is a synthetic e-commerce business with 24 months of orders, returns, inventory, purchase orders and marketing spend behind it. Pick a role from a dropdown on the sign-in page — owner through viewer — and you land on that role's dashboard with nothing to configure.
It is one shared tenant, so treat anything you save in it as public, and it is generated data: the numbers are realistic, not real.
Start your own workspace
You get a new tenant with zero rows in it. Connect Business Central, NetSuite, Acumatica, Shopify or Amazon — or upload a CSV — and the same dashboards fill in from your own data, with the metric definitions already written.
Nothing is shared with the demo: a workspace sees only its own rows, and that boundary is a database policy rather than a query the application has to remember to write.
What you get
- Governed metrics
194 definitions in one catalog. Every dashboard, export, alert and assistant answer compiles from the same one, so two screens cannot disagree about a number.
- Persona dashboards
10 role dashboards — CFO, supply chain, fulfillment, merchandising — rendered from declarative specs, so a role's home screen is data, not a code path.
- The assistant
Ask in English. It plans a validated query against the catalog and never writes SQL, so it cannot invent a table or reach data you could not already see.