Create a Meeting Agenda with AI: Decisions and Timeboxes

Create a meeting agenda with AI by starting with the decisions or outputs the meeting must produce. Provide the participants, available time and required preparation. Ask the model to assign realistic timeboxes and identify topics that can be handled asynchronously. Review the agenda with the decision owner before sending it.

A useful agenda tells people why they are there, what they need to read and what will be different when the meeting ends. A list of broad topics such as “updates” and “discussion” rarely provides that clarity.

Decide whether a meeting is needed#

Write the desired output in one sentence. “Choose one of two launch dates based on the release evidence” is a decision. “Share the release status” may be an update that can be read without a meeting.

If a decision needs discussion, identify who has authority to make it and who supplies relevant information. Do not ask AI to assign decision rights from job titles alone. Provide the actual ownership and note any person whose absence prevents the decision.

Use a realistic 30-minute example#

The following fictional agenda is for choosing a launch date. It assumes the evidence packet has already been read. The times add to 30 minutes; they are a planning example, not a claim that every launch decision fits this schedule.

If the readiness issues cannot be explained in seven minutes, reduce the scope or provide better prework. Asking AI to squeeze more topics into the same time does not create additional capacity.

Copy this agenda prompt#

Create an agenda for a [duration]-minute meeting.
Required decision or artifact: [specific output].
Participants, decision rights and missing roles: [details].
Evidence to review beforehand: [links or summaries].
Constraints, options and open questions: [paste].
Return a timed agenda whose durations add to the meeting length,
prework, an owner and expected output for each item, and a parking-lot
rule. Flag missing decision authority or evidence. Suggest asynchronous
alternatives for updates. Do not invent agreements or availability.

Check the arithmetic and the sequence. The meeting should not ask participants to choose an option before they have reviewed the constraints. Put the most important decision early enough that it receives the time it needs.

Make prework short and specific#

Tell people what to read and what question to consider. A large attachment with no instruction can create more confusion than preparation. Summarize the decision, link to the evidence and identify which information is still uncertain.

If prework is essential, confirm that participants have access. Do not fill the agenda with live reading because a document was never shared. If the evidence changes before the meeting, identify the revision so people do not discuss different versions.

Give each agenda item an output#

Replace “Discuss onboarding” with “Choose the two onboarding steps to test next.” The output can be a decision, a prioritized list, an owner assignment or a clearly defined question for further investigation.

For uncertain decisions, define what a responsible deferral looks like. Name the missing evidence, owner and next decision point. “Discuss again next week” is less useful than identifying exactly what must be learned before the next discussion.

Plan facilitation and participation#

Allow a way for participants to contribute before the most vocal person sets the frame. Depending on the task, that could mean brief silent review, collecting options in advance or inviting the owner of each evidence area to speak.

Ask AI to identify likely agenda bottlenecks, but verify them with the actual group. A cross-team disagreement may need more context than a model can infer from a short brief. The agenda should support a useful conversation rather than pretend to eliminate uncertainty.

Capture decisions separately from notes#

At the end, record the decision, rationale, conditions and action owners. Do not label a suggestion as an agreement. If the decision depends on a later check, preserve that condition in the record.

The next meeting can begin by reviewing unresolved actions rather than repeating the same discussion. Keep the agenda and decision record together so an absent stakeholder can understand what happened without reading a verbatim transcript.

For a concise follow-up, use the project status report template. When the meeting identifies uncertain future events, add them to a project risk register. If a new request changes the deliverables, use the scope-of-work workflow.

Common questions#

Can AI decide who needs to attend?#

It can suggest roles based on the task, but actual decision rights and availability come from the team. Include those facts and review the participant list with the meeting owner.

What if the agenda does not fit the available time?#

Reduce the scope, move updates to prework or schedule a separate session for another decision. Smaller timeboxes do not make complex discussion automatically shorter.

Should every meeting end with a decision?#

Some meetings produce learning or a shared artifact. Define that output clearly; if a decision is deferred, record the missing evidence and the next responsible step.

Download the working files#

Download the copyable prompt and download the illustrative worksheet (CSV). Replace example data with your own verified inputs; the worksheet is a planning aid.

Browse the English AI workflow guides for related tasks.