Three colleagues with laptops at a café table

Launch a training to the right people

The audience built from roles and attributes, assigned when you approve.

For
L&D
Routine
When needed
Setup
Juno + 1 tool
You approve
every draft

The prompt

Written to ask before it assumes, check what already exists, and stop before anything goes live.

Works in Claude, ChatGPT, Copilot
Launch a training to the right people:

Get one training in front of the right people in Juno: assigned, with a due date if we need one, and announced once. Nothing sends until I confirm it from the review card.

═══ ONE ROUND OF QUESTIONS ═══
Ask me in ONE message, then wait:
1. Which training. Find it by name (search_lms_content) and confirm the match with me.
2. Who gets it: everyone, or described by department, location, role, manager, start date, or a pasted list of names or emails.
3. A due date, or none.
4. Whether the built-in assignment email is enough, or whether I want a separate announcement as well. Default to one message, not two.
5. What I mean by "send": assign (it lands on their to-do list and is tracked), notify (a one-off message, nobody is enrolled), or file into a channel (shelf placement). If I am unsure, explain the three in a line each and let me pick.

In that same first message, also ask me my role and the team I support, and which HR system we use. Then check each one is connected in this chat; if not, I'll connect it or upload an export. Never guess what is in it.

═══ CHECK IT IS READY ═══
Confirm the training is published; drafts are invisible to learners, and you never assign or announce a draft. If it is still a draft, validate it, show me, and publish only on my explicit yes. Interactive lessons are live the moment they are saved and need no publish.

═══ WHO EXACTLY, THEN STOP ═══
Resolve the audience (resolve_cohort, or search_users for named people) and show me, in plain words: the total count, a named sample of up to eight people, and whether anyone I named could not be found. Never drop a name silently. If the audience is larger than the platform can send to at once, say so and propose splitting it by a natural group. Any edit to the audience means you resolve and show it again.

═══ SEND FROM THE CARD ═══
Open the assignment for review (prepare_assignment) with the training, the audience, and the due date filled in. I review and send from there; you never send on your own. If I asked for an announcement too, draft it: two to four sentences, what it is, why it matters, the deadline, how long it takes, using only the personalisation fields the platform supports, and open it for review the same way (prepare_notification). Afterwards, report plainly what happened: enrolled, already assigned, invites queued.

Rules throughout: published training only; audience shown before anything opens; every send fires from a card I confirm; one message by default; neutral wording, never singling anyone out; talk in product terms, not tool names.

Your AI asks you

  1. Q1Which training. Find it by name (search_lms_content) and confirm the match with me.
  2. Q2Who gets it: everyone, or described by department, location, role, manager, start date, or a pasted list of names or emails.
  3. Q3A due date, or none.
  4. Q4Whether the built-in assignment email is enough, or whether I want a separate announcement as well. Default to one message, not two.
  5. Q5What I mean by "send": assign (it lands on their to-do list and is tracked), notify (a one-off message, nobody is enrolled), or file into a channel (shelf placement). If I am unsure, explain the three in a line each and let me pick.

The routine

  1. 01

    Connect

    Juno and the tools below

  2. 02

    Answer

    the questions, one message

  3. 03

    Review

    the draft or the view it builds

  4. 04

    You approve

    before anything goes live

What it connects

  • JunoRequired
  • HRISSCIM

Your AI assistant reads the tools it can reach and drives Juno.

Where this Journey Play works