Delivering Results through Joule Skills

A Joule Skill is the smallest, most focused asset you create in Joule Studio. It performs one predefined action—such as creating a ServiceNow ticket or fetching a customer’s open invoices—using fixed inputs and outputs. Because the execution path never branches, every run is predictable, auditable, and fast.

Lots of "micro-requests" still force employees to juggle multiple screens: checking warehouse stock, downloading a payslip, raising a freight claim. By wrapping each of those tasks as a Skill, you let people ask in plain English while the system completes the required API call in the background. Users save time, and compliance teams still get full traceability because every invocation is logged with caller, payload, and result.

The authoring flow is intentionally lightweight:

1. Name your Skill
   • Give it a clear, business-focused name.
   • Write a one-sentence purpose statement for the intent directory.
2. Configure inputs & outputs
   • Drag & drop input and output fields onto the canvas.
   • Map them to the required data.
3. Embed automations
   • Configure automated steps.
4. Add actions & messages
   • Connect fields to SAP Build Actions or third-party connectors.
   • Define the messages the Skill will return.
5. Test & refine
   • Use the built-in simulator to run test payloads.
   • Check the step-by-step trace and adjust mappings as needed.
6. Deploy your skill
   • Publish to a sandbox for sign-off.
   • Promote to production when ready.
   • Each conversation stays pinned to its original version for stability.

Imagine a maintenance planner typing:

Joule recognises the intent to create a purchase requisition and calls the corresponding Skill. The Skill extracts the item, quantity, cost centre, and priority; invokes the Purchase Requisition API in SAP S/4HANA; and replies:

One plain-language request becomes a fully governed transaction—no manual data entry, no context switching.

Quality is enforced much like any other enterprise service.

Teams store common request variants as regression packs and rerun them after each change. When a Skill evolves—perhaps because an extra parameter is needed—developers release version 1.1; if the data schema must break, they publish version 2.0 and leave the earlier branch untouched until all consumers have migrated. Telemetry on latency and error codes highlights Skills that need optimisation or refactoring, and life-cycle tooling lets you freeze or retire older versions once they are no longer required.

Organisations usually begin with high-volume, low-variance operations: retrieving HR policy documents, submitting IT help-desk tickets, or tracking a DHL shipment. These scenarios map cleanly to the deterministic nature of Skills and deliver an immediate productivity boost without changing core business processes.

Joule Skills convert a single user intent into a secure, repeatable system action. They blend conversational ease with enterprise-grade control and form the foundation on which more sophisticated AI Agents can later build.

• Creating a Joule skill with Joule studio