A sample assistant for IBM Supply Chain Intelligence Suite (SCIS)
A team of supply chain analysts needs to review the status of the inventory for the product categories they work on, N95 masks, swabs or syringes for example, typically on a daily basis.
They develop a new Watson Assistant skill for the Sterling AI Assistant that brings up that status when they say something like "Show me the daily status for Pharma" for example.
The list of inventory for a product category is already implemented as a UIHub layout in the SCIS. The new skill simply reuses that UIHub layout, providing convenient access to it from natural language.
This sample project illustrates how to create the Watson Assistant skill, train it with example utterances, recognize the product category mentioned in an utterance, and have it return the specification of the UIHub layout presenting the inventory data that the team wants to see.
The project also contains example test programs that illustrate how to test that the skill accurately recognizes the user's intent and returns the expected UIHub layout.
Documentation of how to design, build and deploy Watson Assistant skills for the Sterling AI assistant can be found in the Sterling Knowledge Center.
Before testing the skill you need to upload it to Watson Assistant.
Please do the following:
-
Create a Watson Assistant service resource in an IBM cloud account.
-
In the Watson Assistant tool, select Create assistant and create an assistant named Daily Report Sample, which will be necessary to host the skill.
-
Write down the assistant URL and API key for the Daily Report Sample assistant, found on that assistant's Settings page.
-
In the Watson Assistant tool, select Create skill, choose Dialog Skill, then select Upload skill to upload
src/skill.jsoninto the new skill. -
Write down the Legacy v1 workspace URL for the Daily Report Sample skill, found on the skill's API details page.
-
Open the Daily Report Sample assistant again, select Add dialog skill to add the Daily Report Sample skill to the assistant.
The project contains a set of Node.js sample test programs which can be used to test the skill.
To build the test programs, make sure you have Node.js and yarn installed, then run the following command:
yarn build
The sample test programs illustrate how to test that the skill functions as expected and with good language recognition accuracy.
Configure the following environment variables related to Watson Assistant:
export ASSISTANT="the assistant URL for the Daily Report Sample assistant"
export KEY="the API key for the Daily Report Sample assistant"
export SKILL="the legacy v1 workspace URL for the Daily Report Sample skill"
The assistant URL and API key can be found on the Settings page of the Daily Report Sample assistant.
The skill's Legacy v1 workspace URL can be found on the API details page of the Daily Report Sample skill.
Since this sample uses the Named Entity Recognition API, you will need to authenticate using a JSON Web Token which can be obtained from the Saascore Authentication API.
Configure the following environment variables related to authentication:
export SCIS_ORGANIZATION="the ID for your SCIS organization"
export INFOHUB_TENANT="the Infohub tenant ID associated with your SCIS organization"
export SAASCORE_BASIC_USERNAME="the username (IBMID) of the authenticating user"
export SAASCORE_BASIC_PASSWORD="the corresponding password of the authenticating user"
Run the following command:
yarn test
Note: certain test cases assume that specific entities are contained in your organization's business data, e.g. 'Pharma' as short for Pharmaceutical in the test question "Show me the evening status for Pharma". You may simply replace this value with a value from your own organization's business data.
To register the sample skill with the Sterling AI Assistant, follow the steps described in Registering your skills with the Sterling Assistant.