- An AI session for a sales meeting works best as live demos on the sales cycle the team already runs: account research, account planning, meeting prep, objection handling, follow-ups and proposal drafts, in the AI tool the company already approved.
- For industrial and technical B2B sales, every demo needs an accuracy check: specs, compatibility claims, prices and commitments come from approved documents and a person, never from the AI.
- In a large room, demos run on one shared screen and reps practise in pairs that mix an experienced AI user with a newer one, so different experience levels move together.
- Ground rules come first: which tools are approved, what never goes in (pricing sheets, contract terms, customer personal data), and that a rep reads and sends every message.
- The session ends with each rep choosing one workflow to use on a real account the next week, with a template to start from.
What a sales meeting AI session should cover
An AI session for a sales meeting should show reps how to use the AI tool the company already approved on the sales cycle they already run: research an account, plan it, prepare for the meeting, handle objections, follow up and draft a proposal. It should be live demos on realistic sales material, with an accuracy check in every segment, not an hour of prompt basics.
Requests for this session tend to be specific. A sales leader has a slot at the next team meeting or kickoff, the room is a few dozen reps with very different AI habits, and the brief is "show us how to use it in our sales process." Industrial and technical sellers add two conditions: the cycle is long and runs on relationships, and a wrong spec in an email does real damage. This agenda is built for that brief. For training built on each rep's own accounts, see our AI training for sales teams.
The 90-minute agenda
A 90-minute AI session for a B2B sales meeting runs in six blocks: ground rules, account research and planning, meeting prep with objection practice, follow-ups and LinkedIn, proposal first drafts, and a close where each rep picks one workflow for next week. Each block is a live demo followed by a short check of what the AI got wrong.
| Minutes | Block | What the room sees |
|---|---|---|
| 0–10 | Ground rules | Which tools are approved, what never goes in, and that a rep sends every message. |
| 10–30 | Account research and planning | A one-page account brief from public pages and sample CRM notes, a source after every line, and a list of unknowns. |
| 30–50 | Meeting prep and objections | Discovery questions and an agenda from the brief, then the AI plays a skeptical buyer while a rep practises answers. |
| 50–65 | Follow-ups and LinkedIn | A follow-up drafted from call notes, with one invented commitment the room has to catch. |
| 65–80 | Proposal first drafts | A first draft assembled from an approved past proposal and the buyer's own words, with the technical check marked. |
| 80–90 | Pick one workflow | Each rep chooses one workflow and one real account to use it on next week. |
With 120 minutes, keep the same blocks and add 30 minutes of paired practice after the follow-up block, so every rep runs one workflow on their own laptop before leaving the room.
Account research and account planning
Start with account research because it is the most common sales task and the easiest to check. Live, on the shared screen, the AI turns a target company's website, its published product pages and a few lines of sample CRM notes into a one-page brief. Every line names its source, and anything the sources do not say goes on a list of unknowns.
For technical and industrial accounts, the brief should separate three things: what the company makes and sells, who is likely involved in a purchase (engineering, operations, procurement), and what changed recently, such as a new site or a published project. Then the demo moves from one contact to the account plan: the same sources, rewritten as open questions for each stakeholder and a proposed next step for each. The prospect research workflow and its template are the version reps can copy afterwards.
The check is the point of the block. Pick two facts from the brief and have the room find them in the source. If one cannot be found, the AI filled a gap, and that line comes out.
Meeting prep and objection practice
Meeting prep turns the account brief into a short agenda, three to five discovery questions and the objections the buyer is likely to raise. The useful part for a live room is role-play: the AI plays a skeptical plant manager or procurement lead, and a volunteer rep answers in front of everyone.
The sales meeting preparation workflow covers the prep itself. Set the role-play up with the buyer's real context: their role, what they care about, what the last call covered. Then ask the AI to push back the way that buyer would, one objection at a time. Reps hear their own answers out loud and the room suggests better ones. This is practice, not a script. In relationship selling the rep still decides what to say and when, and the AI's objections are guesses, not research on the real buyer.
Follow-ups, LinkedIn and the invented commitment
A follow-up drafted by AI is only useful if it matches what was actually said. In this block the AI drafts a follow-up email from sample call notes, in the rep's voice, with what was agreed and what happens next. Then the room finds the line the AI made up, because a convincing follow-up can still promise something nobody agreed to.
The demo uses a deliberately flawed draft. A line like "Thanks for approving the budget; we'll start the rollout next Monday" reads well, but neither the approval nor the date is in the notes. A second version adds "please send the sample by Friday", a deadline the buyer never agreed to. Reps learn to separate four things in every draft: confirmed facts, unknowns, questions and proposed next steps. The sales follow-up workflow and the follow-up email template carry that check.
LinkedIn messages follow the same rule. Personalise from something the person actually published or said, and leave out anything that pretends to a familiarity the rep does not have.
Proposal drafts and technical accuracy
AI can assemble a proposal first draft from an approved past proposal, the call notes and the buyer's own words. In technical B2B sales, it must not supply specifications, compatibility claims, delivery dates or prices. Those come from approved documents and the people who own them, and the draft marks every place where they need to check.
Show this on screen: ask for a draft that leaves placeholders such as [SPEC: confirm with engineering] instead of numbers. Then ask the same question without that instruction and let the room see the AI fill in plausible numbers. That contrast teaches more than a rule on a slide.
Ground rules for confidentiality and accuracy
Before any demo, agree three answers with the sales leader: which AI tools are approved for company work, which information never goes in, and who checks output before it reaches a customer. Write them on one page the reps keep. Pricing sheets, contract terms and customer personal data normally sit on the never list.
Use company workspaces for company work. OpenAI's business data page states that it does not train its models on data from its business products by default; personal accounts are a different matter. Your IT or admin team decides what is approved. The free AI data boundaries one-pager is a starting point for the one-page rules.
How to handle a room with mixed AI experience
A sales meeting usually mixes reps who use AI every day with reps who have barely opened it. Run every demo on one shared screen, pair an experienced user with a newer one for the practice moments, and give everyone the same copyable template, so beginners follow the steps and experienced reps save the workflow in a Project.
Use one tool for the demos if the company has standardised on one, whether Microsoft Copilot, ChatGPT or Claude. The workflows are the same across tools, so reps on another approved tool can follow along. Ask the room to bring laptops, and have the sample material ready to paste so nobody spends the session looking for files.
When this does not work
A sales meeting AI session does not work without an approved tool. If reps can only use personal accounts, the session teaches them to put company work there, which is the opposite of what a sales leader wants from the meeting.
Approve a tool first; the AI tool approval checklist is built for that decision.
It does not work as a substitute for clean source material. If CRM notes are thin or out of date, the AI produces a tidy brief of bad information. And it is the wrong session if the goal is software: connecting AI to the CRM so it reads and writes records on its own is a separate AI workflow build, not a meeting segment.
It also does not replace practice on real accounts. A 90-minute session shows the method; reps get good at it by using one workflow on their own accounts the following weeks.
How to book the session
Jack Qian runs this session himself, onsite in Dallas–Fort Worth or live online, in English or Chinese. Tell him the meeting date, the city, how many reps will be in the room and which AI tools are approved; group size and length are agreed in a written quote.
The session can be scoped as a 90-minute team workshop inside your meeting agenda, or as part of a full-day onsite training.
Jack has taught the same method in community rooms: 309 people registered for his 6 AI events, five in Dallas and Plano and one online. To plan the AI block of your next sales meeting, book a 20-minute call or request a written quote.
The approach in this post is what we teach in AI training for sales teams.
Questions people ask
How long should an AI session at a sales meeting be?
Ninety minutes covers six blocks: ground rules, account research and planning, meeting prep with objection practice, follow-ups, proposal drafts, and picking one workflow for next week. With 120 minutes, add 30 minutes of paired practice so every rep runs one workflow on their own laptop.
Is it a lecture on prompt basics?
No. It is live demos on the sales cycle the team already runs, each followed by a check of what the AI got wrong. Prompt wording is taught inside the demos, as part of building a workflow a rep can repeat.
Which AI tool should we use for the demos?
The one your company already approved for work, usually Microsoft Copilot, ChatGPT or Claude. The workflows are the same across tools, so reps on another approved tool can follow along.
How do you keep customer information safe in a live demo?
Demos use public pages and synthetic or redacted sample notes. Before the session the sales leader confirms which tools are approved and what never goes in, such as pricing sheets, contract terms and customer personal data. A rep reads and sends every message.
Can AI write technical specs or quotes for our proposals?
No. In the session, AI drafts the structure and the wording around the buyer's needs, and leaves marked placeholders for specs, compatibility, dates and prices. Those come from approved documents and the people who own them.
How is the session priced?
Choose a 90-minute team workshop, full-day onsite training, or 4 live 90-minute sessions. Every format is priced by a scoped quote. Group size, location and preparation are agreed in the written quote.