Your supply chain data sits in ERPs, planning suites, warehouse and transport systems, service tools, lakes, databases, files and web services.
One translation layer per kind of system; everything above it works the same way.
Each kind of system publishes its own way in. SAP publishes released services and views for integration. Oracle Fusion publishes REST resources. A warehouse system publishes REST calls; a lake answers SQL; a supplier portal drops a file. A connector holds the settings for one of those kinds.
When a data set is registered, the connector asks the system what it holds and hands the description to the data fabric. When a question arrives, the federated query engine plans the work, sends a query to each source that holds part of the answer, and joins the results in the middle.
The layers above the connector stay the same. ChatLima, an agent, a SmartApp and a BI tool all ask in the words of the ontology, so a question about open purchase order lines reads the same whether the lines sit in SAP, in Oracle or in both.
The layer below stays the vendor’s. Juno uses the routes each vendor publishes and supports, so an upgrade of the source system keeps working. Write-back follows the same principle, through the interface the owning system supports.
Nine categories cover the estates we meet; the general connectors take care of the rest.
| Category | Examples | What Juno reads | How it is read |
|---|---|---|---|
| ERP | SAP S/4HANA · SAP ECC · Oracle Fusion · Oracle E-Business Suite | Purchase orders and lines, sales orders, deliveries, stock, materials, suppliers, planned orders | Released services and views; tables where your team allows it |
| Planning suites | Demand, supply and inventory planning systems | Forecasts, planned orders, exception messages, safety stock and inventory targets | REST resources and published extracts |
| Warehouse and transport | Warehouse management, transport management, carrier portals | Stock by location, tasks, handling units, shipments, deliveries | REST resources, files |
| CRM and service | Salesforce · ServiceNow | Customers, orders, cases, incidents and the tasks raised against them | REST over the system’s own object model |
| Lakes and cloud warehouses | Cloud data warehouses and lakehouses | History at scale, harmonised extracts, measures prepared by your data team | SQL through the warehouse’s own query service |
| Databases | PostgreSQL · Oracle · SQL Server · MariaDB · MySQL | Any table or view the service account may read | SQL |
| Files | CSV, Excel, XML and JSON on a share, in object storage or over SFTP | Supplier confirmations, price lists, uploads from the business | A drop on a schedule, or a direct upload |
| Streams and events | Message brokers and event topics | Movements, confirmations and status changes as they happen | Subscribe and land as a data set |
| Web services | Supplier portals, marketplaces, internal services | Whatever the service publishes, with its own authentication | REST, called on a schedule or on demand |
The full list with versions and settings lives in the connector list. Product.
Juno reads through the routes each vendor publishes, so an upgrade of the source keeps the connection working.
| System | How Juno reads it | Objects your team will recognise | What to agree with your team |
|---|---|---|---|
| SAP S/4HANA · cloud editions | Released OData services and the views published for integration | Purchase order and schedule line, supplier confirmation, planned order, production order, stock, business partner | Which communication scenarios the service account holds |
| SAP S/4HANA private edition · SAP ECC | The same released services, plus direct reads of tables and views where your team allows them | EKKO · EKPO · EKET · EKES · MARC · MARD · LFA1 · PLAF | The client, and whether direct table reads are allowed |
| Oracle Fusion Cloud | REST resources under the Fusion REST interface, plus the extracts your administrator publishes | purchaseOrders · supplyPlans · itemsV2 · inventoryOnhandBalances · suppliers | Which roles the integration user holds |
| Oracle E-Business Suite | Direct reads of the application tables and views your database team exposes | PO_HEADERS_ALL · PO_LINES_ALL · MTL_SYSTEM_ITEMS | The schema, and which views are opened |
/sap/opu/odata4/sap/api_purchaseorder_2/SAP S/4HANA: the released purchase order service in OData V4. It succeeds the V2 service, which SAP deprecated with release 2308.I_POSupplierConfirmationAPI01SAP S/4HANA: the released view behind supplier confirmations, which became a business object of its own with release 2111./fscmRestApi/resources/11.13.18.05/purchaseOrdersOracle Fusion: the purchase order REST resource. Drafts and change orders carry their own resources beside it.Read paths shown for orientation; your estate decides which of them is switched on. Product · dated vendor documentation, checked 19 September 2026.
The mode belongs to the data set, so one project can mix all three.
The mode is set per data set and can change later. Product.
Showcase estate.
The showcase estate holds SAP S/4HANA for the main business, Oracle Fusion for an acquired division, a planning suite, a supplier portal and a lake. Four routes in were needed before L. Tan could ask her question.
SAP answered through its released services and the purchasing views. Oracle Fusion answered through its REST resources once S. Iyer granted the integration user the purchasing roles. The planning suite gave a published extract of planned orders and exceptions. The supplier portal dropped a daily file of confirmed ship dates at 06:00. All four arrived in the same project, with one description each.
How one question becomes a query per source, how the results are joined, and when a governed copy is the better choice.
SAP S/4HANA in its cloud and private editions, and SAP ECC. Cloud editions are read through released services and published views; the private edition and ECC add direct reads of tables and views where your team allows them.
Anything with a database, a REST interface, a file drop or a message topic can be registered through the general connectors. The named connectors add ready settings for systems we meet often.
Agents write through the interface the owning system supports, switched on per application after a confirmed test write. The system runs its own checks and its acknowledgement is kept with the action.
More than 40 source systems are connected across production estates, with 60 GB under management and over 200 active business users.
Each one builds on the one before it.
Reads your rows in place.
Connect and move ↗One record per real thing.
Master and govern ↗Apps for every decision.
Operate and decide ↗Asks and acts, at your level.
Ask and act ↗Bring two systems and one decision your planners make every week. In the first week we register the data, confirm the mapping with your team, answer your questions in ChatLima and run the first agent at the level you choose.