myWork Learn: authoring and rolling out
myWork Learn keeps your team current. The AI drafts short, role-tailored questions from your Knowledge Base; each person gets a couple of them as a daily check; and the results feed a low-stakes, positive KPI signal (never a gate). It is gated by the LEARNING_AUTHOR permission and two feature flags.
The lifecycle
1. Create a topic
Open People > Learning > Review & topics, then the Topics tab. Add a topic: give it a title, link the KB article slug(s) the questions should draw on, and add competency tags.
2. Get questions (AI authoring)
The external AI generator pulls a prompt-pack for the topic (the writing policy, the linked KB source text, and the existing questions so it does not repeat itself) and submits multiple-choice proposals.
- Well-formed, high-confidence proposals auto-publish.
- The rest land in the Review queue for a subject expert to approve, edit the wording, or reject.
- Only approved questions are ever served to learners.
Authoring writes require the appService.learningAuthoring flag to be on.
3. Activate the topic
A new topic is a Draft. On the Topics tab, click Activate (it unlocks once the topic has at least one approved question). Only Active topics are served. Click Retire any time to stop serving without deleting anything.
4. Assign the audience
Click Audience on the topic and add one or more targets: a job position, a department (optionally including its sub-departments), or a whole organization. Only people who match the audience will get the topic.
5. It is served, and feeds KPI
Everyone in the audience now gets a couple of the topic's questions in their daily check. Answers are graded instantly; questions people get wrong come back sooner (spaced repetition). Attempts feed a low-stakes positive learning signal on the person's KPI.
Settings
- Cadence (questions per day, minimum gap, re-ask timing, rest interval) is configured in App Config, group
Learning- no redeploy needed. - Two flags, both off by default:
appService.learningAuthoring(the AI authoring writes) andappService.learningModule(serving the daily check to learners). - Difficulty to points: Recall = 1, Apply = 2, Analyse = 3 (harder questions are worth more).
Good to know
- Nothing reaches a learner unless the question is approved and the topic is active with an audience.
- A person with no matching audience simply gets nothing; it is opt-in by role.
- Free-text answers are captured but not graded yet (multiple-choice only for now).
- When a linked KB article changes, its questions can be flagged for re-review so the check stays accurate.