Choose a follow-up worth repeating

The artifact for this moment is a bounded request with a named artifact, reader, and decision owner. That description matters because “help with choosing the follow-up task” is a conversation starter, not a quality standard. A useful request identifies the reader or user, the available material, the constraint that would make the answer unusable, and the decision that remains outside the chat. Missing facts should appear as questions or [UNKNOWN], never as invitations to invent a believable answer.

For choosing the follow-up task, Riley, improving meeting-note follow-ups in a thirty-minute weekly loop can use a compact working note:

JOB: produce a reviewable draft for choosing the follow-up task
KNOWN: use only confirmed details from this week's work
CONSTRAINT: make uncertainty visible; do not add names, dates, or promises
HUMAN DECISION: approve, verify, or decline the next action

Notice the decision before asking for language. Ask what would change if one fact were reversed. If changing a number, access need, deadline, source, or approval changes the artifact, that fact belongs in the request. If it does not, leave it out. This test keeps choosing the follow-up task connected to the week's actual work rather than turning into a generic example.

Riley keeps the tool fixed and changes one variable this week, so any improvement has a traceable cause. In particular, do not treat a draft as authorization. Check original notices, labels, course materials, household agreements, or team records when they are the source of truth. Escalate to a qualified person when safety, health, legal obligations, accessibility, financial commitments, or another person's consent is involved.

This boundary is useful even when the task is low stakes. It lets Riley, improving meeting-note follow-ups in a thirty-minute weekly loop move quickly on formatting and options while preserving time for the parts that require local knowledge and responsibility.

That narrow move creates evidence about the conversation. If the second version still crosses the boundary, stop using the output for that purpose and complete the decision with the appropriate person or source. The useful next step after choosing the follow-up task is recording a baseline; carrying an uncorrected assumption forward would make the next page harder.