Using AI for meeting preparation.

Twenty minutes of preparation with AI can change how an important meeting goes. Account managers, owners, project leads and team leaders will leave this lesson with a repeatable routine that turns material they already have into a one-page brief they trust, plus a short rehearsal of the questions they would rather not be asked. You need an upcoming meeting, the notes and emails behind it, and any AI assistant.

What a good meeting brief contains

A meeting brief is one page that tells you why you are in the room and what has to happen before you leave. The useful sections are:

The split in the last column is the lesson. AI is good at organising and suggesting; the objective and limits are judgement calls that should never be delegated.

Exercise: brief your next meeting

1. Pick a meeting that matters. One in the next week where you need a decision, not a catch-up.
2. Gather three to five inputs. Last meeting’s notes, the recent email thread, the key numbers from your own systems and any agenda already sent. Remove personal details you do not need.
3. Write the objective and limits yourself. One sentence each, before you open the AI tool.
4. Run the brief prompt. Use the worked example below as a pattern. Ask for gaps to be flagged rather than filled.
5. Check “What we know” line by line. Each fact should match its source. Delete anything you cannot trace.
6. Rehearse. Ask the assistant to play the other side and put three hard questions to you. Write a one-line answer to each.

Worked example: a Marlborough winery’s buyer meeting

Illustrative example: the sales manager of a mid-sized Marlborough winery is meeting the beverage buyer of a Wellington restaurant group to renew a by-the-glass listing, with a 6% price increase on the table. She pastes her material into the business AI account:

Prepare a one-page meeting brief from the material below.

My objective: renew our Sauvignon Blanc by-the-glass listing across their four
venues for 12 months, at the new price.
My limits: no discount below the new price; I can offer staff training visits
and a later price start date of 1 January.

Sections: What we know (each fact with its source in brackets); Agenda (30 minutes,
with times); Decisions needed; Questions to ask; Likely objections with a suggested
response; Gaps (anything the material does not tell me).
Use only the material below. Do not add facts about the restaurant group.

MATERIAL
[last meeting notes, March]
[email thread summary, September]
[our depletion figures for their venues, last 12 months, from our sales system]
[new price list]

Example output (illustrative and abridged):

WHAT WE KNOW
- Their four venues poured our Sauvignon Blanc all year; volume dipped in winter
  at the waterfront venue (depletion figures).
- In March they asked for staff training on food matching (March notes).
- Our increase is 6% from 1 November (price list).

DECISIONS NEEDED
- Renewal for 12 months at the new price, and the start date.

LIKELY OBJECTIONS
- "Competitors held their prices." Response: offer the 1 January start and two
  staff training visits, both within your limits.

GAPS
- The material does not say whether they are reviewing all wine suppliers or only us.

The gap is the most useful line: it becomes her first question in the meeting.

How to rehearse the hard questions

Rehearsal is where AI earns its keep for meeting preparation, because it will ask the awkward question you have been avoiding. After the brief is checked, stay in the same chat and write: “Play the buyer. Using only the material above, ask me the three hardest questions you would ask in this meeting, one at a time. After each of my answers, tell me in one sentence what a sceptical buyer would still doubt.”

Answer out loud, then type a one-line version of each answer into the brief. Two rounds is usually enough. If a question exposes something you genuinely do not know, add it to Gaps and find out before the meeting rather than improvising. Teams that prepare for important meetings together often practise this routine in our AI training for business teams.

Where AI meeting preparation goes wrong

• Invented background. Asked about the other organisation without material, assistants produce plausible history, such as an expansion that never happened. Supply sources, or verify using checking AI answers against sources.
• AI arithmetic. Calculate numbers in your own systems and paste the results. Do not ask the assistant to total sales figures.
• Researching people’s lives. The Privacy Commissioner’s guidance treats asking AI to generate information about a person as collecting personal information, even when the result is wrong (OPC). Stick to the organisation and professional roles. The lesson on researching a customer account covers this properly.
• Confidential limits in the wrong tool. Price floors and negotiating positions are confidential business information. Use your business account, not a personal one.
• Scripts instead of listening. A brief is for walking in prepared, not for reading out.

Tools that sit inside your email and calendar can save the gathering step. Microsoft says a paid Copilot licence adds Copilot inside Outlook and Teams with grounding in your organisation’s data, while free Copilot Chat is primarily web grounded (Microsoft, Microsoft Learn). Google lists Gemini in Gmail, Docs and Meet from Workspace Business Standard, as at 19 October 2026 (Google). Our guides to Microsoft Copilot, AI and Microsoft Teams and AI and Google Workspace cover set-up, and the research briefing preparation workflow shows the same method for professional services firms.

Check your brief

• I wrote the objective and limits myself before using AI.
• The brief fits on one page.
• Every line in “What we know” has a source I checked.
• Numbers came from my own systems, not from the AI.
• Research on the other party stays at the organisation and role level.
• I rehearsed three hard questions and have an answer for each.

Next lesson: Turning meeting notes into actions covers what happens after the meeting.

Written by the Levy Street team in Wellington. Last updated 19 October 2026.