Run Meetings That Drive Decisions

Running Better Meetings with AI

It's 4:47 PM on a Thursday. You just wrapped a 90-minute project status meeting. Eight people were in the room. Three of them spent half the time on their laptops. Someone asked a question in the last ten minutes that was already answered in the first twenty. And now you're staring at a blank document, trying to remember who agreed to do what by when, because nobody wrote it down clearly. You need to send a recap email before end of day. You have no notes. You have a headache. This is not a rare situation. For most project managers, team leads, and department heads, this is Tuesday.

What AI Actually Does for Meetings

AI tools work in three phases of any meeting: before, during, and after. Most professionals start by using AI after the meeting, writing summaries, drafting follow-up emails, turning notes into action items. That's a solid starting point. But the bigger gains come when you build AI into all three phases. Before a meeting, AI helps you design sharper agendas, anticipate blockers, and prepare talking points for difficult conversations. During a meeting, AI transcription and note-taking tools capture everything in real time so you can be fully present. After a meeting, AI converts raw transcripts into structured summaries, assigns action items, and drafts communications, in minutes rather than the 45 minutes most people spend on post-meeting admin.

Think of AI as the most reliable colleague you've never had, one who takes perfect notes every time, never zones out, doesn't have opinions about whose idea was better, and can produce a clean summary in the tone you specify before you've even left the room. The tools that do this aren't experimental. They're in tools you may already pay for: Microsoft Copilot is built into Teams meetings for Microsoft 365 subscribers. Google Gemini integrates with Google Meet. Otter.ai works with Zoom, Teams, and Google Meet. Fireflies.ai joins meetings as a bot and delivers transcripts plus AI summaries to your inbox within minutes of the call ending. These are not futuristic promises, they're features available right now.

Click to enlarge

Phase One: Building Smarter Agendas Before You Walk In

A bad agenda is a list of topics. A good agenda is a sequence of decisions. The difference matters enormously in practice. When an agenda just says 'Q3 budget update,' nobody knows whether they're there to hear information, debate a shortfall, or approve a reallocation. People arrive unprepared, discussions spiral, and you run out of time before reaching the items that actually needed the room. AI is surprisingly good at helping you convert a loose list of topics into a structured decision-making agenda, complete with time allocations, the specific question each agenda item is trying to answer, and the materials people need to review in advance.

The technique is straightforward. Open ChatGPT, Claude, or Microsoft Copilot. Describe your meeting, who's in the room, what project it's about, what happened since the last meeting, and what you need to resolve before the project can move forward. Then ask the AI to draft an agenda structured around decisions rather than topics. You'll get a starting draft in under 60 seconds. Edit it. Add context the AI couldn't know. Send it to your team 24 hours in advance. This single change, agendas framed as decision sequences, consistently reduces meeting length and increases the rate at which meetings end with clear next steps, according to research from Harvard Business Review on meeting effectiveness.

Click to enlarge

Choosing the Right AI Tool for Each Meeting Phase

The AI meeting tool landscape has expanded quickly, and the right choice depends on your existing tech stack, your budget, and how much friction you're willing to tolerate. Some tools are fully integrated into the video platforms you already use. Others are standalone apps that join any call as a bot. A few require nothing more than pasting your notes into a chat window. Here's a practical comparison of the main tools professionals are using right now, what they do well, what they cost, and where they fall short.

If your organization runs on Microsoft 365, Copilot in Teams is the path of least resistance, it's already there, it requires no new logins, and it produces meeting summaries and action item lists without any extra steps. If you're on Google Workspace, Gemini in Meet is the equivalent. If your meetings span multiple platforms or include external clients who wouldn't appreciate a bot joining the call, the manual approach, recording the meeting yourself, then pasting a transcript or rough notes into ChatGPT or Claude, gives you full control with no setup required. The 'right' tool is the one you'll actually use consistently, not the most sophisticated option on the market.

Phase Three: Turning Messy Notes Into Clean Outputs

Post-meeting documentation is where most people feel the pain most acutely. You have 45 minutes of raw conversation, three pages of scribbled notes, and a list of half-finished action items. Turning that into a usable summary, a clear set of tasks, and a follow-up email to stakeholders used to take 45 minutes of focused writing time. With AI, it takes about four minutes. The process is simple: paste your raw notes or a transcript excerpt into ChatGPT or Claude, describe the output you need, and specify the tone and audience. The AI does the heavy lifting. You review, adjust, and send.

The key to getting useful outputs is being specific about what you want. Asking an AI to 'summarize my meeting notes' will give you something generic. Asking it to 'extract all action items with owners and due dates, then write a three-paragraph summary for a senior stakeholder who wasn't in the room' gives you something you can actually use. This specificity is what separates professionals who get consistent value from AI from those who try it twice and give up. The technique is called prompt engineering in technical circles, but for practical purposes, think of it exactly like briefing a very capable new team member. The more context you give, the better the output.

Putting This Into Practice on Monday

The fastest way to build an AI-assisted meeting habit is to attach it to a meeting you already have scheduled. Look at your calendar for next week. Pick one recurring meeting, a team standup, a client check-in, a project status call, and commit to using AI for one specific task connected to that meeting. If you're running the meeting, use ChatGPT or Claude to draft the agenda before you send it. If you're attending the meeting, take rough notes during it and paste them into Claude or ChatGPT immediately afterward to produce a summary and action list. Do this for three consecutive meetings before you add anything else.

The time savings are real and measurable. Most project managers report spending 30–60 minutes per week on post-meeting documentation, summaries, follow-up emails, updating project trackers with new action items. AI reduces that to 10–15 minutes for the same output quality, often better. Over a month, that's two to four hours returned to your schedule. That's not a small number. Over a year, it's the equivalent of three full working days spent on post-meeting admin that you get back for actual project work.

Consistency matters more than perfection here. Your first AI-generated agenda won't be perfect. Your first AI-processed meeting summary will need editing. That's expected, and it's still faster than doing it from scratch. The goal in the first two weeks is to build the reflex: meeting happens, open AI tool, paste notes, get draft, edit and send. Once that reflex is established, you can start refining your prompts, experimenting with different tools, and expanding into the before and during phases. The professionals who get the most value from AI are those who start narrow, build consistency, then expand, not those who try to transform everything in week one.

Click to enlarge

How This Looks Different Across Roles

• Project Managers: Use AI to build decision-focused agendas, automate post-meeting action item lists, and draft status update emails to senior stakeholders, cutting post-meeting admin from 45 minutes to under 15.
• HR Managers: Use AI to prepare structured interview debrief agendas, convert panel feedback notes into candidate comparison summaries, and draft hiring decision communications.
• Sales Managers: Use AI to prep call agendas with client context included, convert call notes into CRM-ready summaries, and draft follow-up proposals or next-step emails while the conversation is still fresh.
• Marketing Managers: Use AI to structure campaign briefing meetings around clear decisions (budget, channels, messaging approval), then convert creative review notes into actionable feedback documents for agencies or designers.
• Consultants: Use AI to prepare client-facing meeting agendas that signal structure and professionalism, then produce polished meeting recaps and next-steps documents that reinforce the value of each engagement.
• Teachers and Training Managers: Use AI to plan structured team or parent meetings with clear objectives, then convert discussion notes into action plans or communication summaries for wider distribution.
• Small Business Owners: Use AI to make every team meeting count, sharper agendas, documented decisions, and follow-up emails that keep staff accountable without requiring a dedicated project coordinator.

Click to enlarge

Key Takeaways From Part One

1. AI works in all three meeting phases, before (agenda), during (transcription), and after (summaries and follow-ups), and you can start with just one.
2. The most immediate time saving for most professionals is in the after phase: turning rough notes into polished summaries and action lists takes minutes, not an hour.
3. Decision-focused agendas, where each item is framed as a question to be answered, consistently produce shorter, more productive meetings than topic-based agendas.
4. The right AI meeting tool depends on your existing tech stack: Copilot for Microsoft 365 users, Gemini for Google Workspace users, Otter.ai or Fireflies.ai for cross-platform needs, and manual ChatGPT or Claude for maximum control.
5. Specific prompts produce useful outputs. Treat AI like a capable new team member being briefed: give context, specify the output format, name the audience, and state the tone.
6. Start with one meeting, build the reflex over two to three weeks, then expand. Consistency beats ambition in the first month of any new AI habit.

From Raw Notes to Real Decisions: AI in the Middle of Your Meeting Workflow

Picture this: it's Thursday afternoon, and you've just wrapped a 90-minute project status meeting with eight stakeholders. Your notes are a mess of half-sentences, someone's action item got buried under a budget tangent, and two people left the room with completely different understandings of the next deadline. By Friday morning, you're still reconstructing what was actually decided. This isn't a time management failure, it's a structural one. The meeting produced decisions, but your system for capturing and distributing them didn't keep up. This is exactly the gap AI closes, and it closes it fast.

Making Sense of Meeting Chaos in Real Time

Most meeting tools now offer AI-assisted transcription and summarisation that runs during the call itself. Microsoft Copilot inside Teams can generate a live summary, flag who spoke the most, and surface action items before the meeting even ends. Otter.ai does the same for Zoom and Google Meet calls. The output isn't perfect, names get confused, crosstalk creates garbled lines, but the core value is enormous: you walk out with a structured draft rather than a blank page. Think of it as having a very fast note-taker who never gets distracted, never forgets to write something down, and can produce a formatted document in under 60 seconds after the call ends.

The bigger shift is what this unlocks for project managers, team leads, and consultants who chair multiple meetings per week. When you're not scrambling to take notes, you can actually listen, ask sharper follow-up questions, notice when someone goes quiet, read the room. The AI handles the administrative layer so you can focus on the human one. That's not a small thing. The quality of your facilitation directly affects the quality of the decisions your team makes, and you can't facilitate well when half your brain is typing. AI note-taking is less about saving time after the meeting and more about improving performance during it.

Turning a Transcript into a Project Artefact

A raw transcript is not a meeting summary. It's 40 pages of 'ums', repeated points, and tangents about the office printer. The real skill, and where AI earns its keep, is transforming that transcript into something your project actually needs: a clean action log, a decision register, a risk flag, or a stakeholder update. You can paste a transcript into ChatGPT or Claude and ask it to extract exactly what you need. The prompt matters here. Vague instructions produce vague outputs. Specific instructions, specifying the format, the audience, and the purpose, produce outputs you can actually use without heavy editing.

Notice what that prompt did. It told the AI the exact output format needed (decisions, action table, risks), it specified the audience (senior stakeholders), and it set the tone (professional, concise). The AI didn't have to guess. Compare that to a prompt like 'summarize this meeting', you'd get a paragraph that mixes everything together and is useful to no one in particular. The format of your request shapes the usefulness of the response. This is prompt engineering, and for non-technical professionals, it simply means: be as specific as you would be when briefing a good assistant.

Choosing the Right AI Tool for Your Meeting Workflow

The market for AI meeting tools has grown quickly, and the options look similar on the surface. They're not. Some tools are built into platforms you already use; others require separate subscriptions and app installs. Some work best for internal team meetings; others are better suited for client calls or sales conversations. The right choice depends on where your meetings happen, who attends them, what you need from the output, and what your organization allows from a data security standpoint. The table below maps the most widely used options against the factors that matter most for project management professionals.

What You Do Differently Starting Monday

The practical change isn't just adding an AI tool to a meeting, it's redesigning your meeting workflow around a three-stage pattern: prepare, capture, distribute. In Part 1, we covered preparation. The capture stage is where AI transcription and live summarisation sit. The distribute stage is where most professionals currently lose the value they've created. Action items get buried in an email thread. Decisions aren't formally recorded. The team leaves the meeting and immediately begins diverging in their understanding of what was agreed. AI fixes this by making the distribute stage nearly automatic, and far more structured than a 'great chat today, here's what we discussed' follow-up.

After a meeting ends, your new workflow looks like this: open the AI-generated transcript or summary, paste it into ChatGPT or Claude with a targeted prompt, and produce three outputs in under ten minutes, a decision log entry, an action item table for your project tracker, and a follow-up email to attendees. Each output serves a different audience and a different purpose. The decision log is for the project record. The action table feeds into whatever tool your team uses. Asana, Monday.com, Jira, or even a shared spreadsheet. The follow-up email closes the loop with attendees and creates a written record that everyone received the same information.

This three-output approach takes roughly the same time as writing a decent follow-up email used to. The difference is you're now producing three targeted documents instead of one general one, and each is formatted for its specific purpose. Over the course of a project, this compounds. At the end of a six-week sprint or a quarterly planning cycle, you have a clean, searchable archive of what was decided, who owned what, and when it was due. That archive is invaluable during project reviews, audits, and, frankly, when disagreements arise about who agreed to what.

Click to enlarge

How This Changes by Role

• Project Managers: Use AI to auto-populate your RAID log (Risks, Assumptions, Issues, Dependencies) after each steering committee meeting, paste the transcript and ask for any new risks or blockers mentioned.
• Marketing Managers: After campaign briefing calls, ask AI to extract the agreed messaging pillars, target audience decisions, and channel priorities into a one-page brief your team can work from immediately.
• HR and People Teams: After interviews or panel debriefs, use AI to structure evaluator feedback against predefined criteria, faster shortlisting without losing nuance from the conversation.
• Sales Professionals: After a discovery call, paste your notes into ChatGPT and ask it to identify the prospect's stated pain points, buying signals, and objections, then use that to draft your follow-up proposal outline.
• Consultants: After client workshops, generate a structured 'what we heard' summary that maps client inputs to your framework, saves hours of post-workshop synthesis and gives clients a fast deliverable.
• Teachers and Trainers: After a curriculum planning meeting or parent conference, use AI to summarize agreed accommodations, action points, and follow-up dates in a format you can file or share with colleagues.
• Executives: Ask AI to produce a one-paragraph decision brief from each board or leadership meeting, who decided what, on what basis, and what happens next. Keeps your personal record clean without extra admin.

Writing Smarter Follow-Ups: The AI Draft That Doesn't Sound Like AI

The follow-up email is where most meeting value evaporates. People write them quickly, skip context, and assume everyone remembers the same things, they don't. A well-structured follow-up email is a project management tool in itself. It confirms decisions, distributes accountability, and gives anyone who missed the meeting a clear picture of where things stand. AI can draft this in seconds, but only if you give it the right inputs. The key is asking for a specific tone and structure, not just 'write a follow-up email.' The prompt below shows exactly how to do this for a project status meeting with mixed seniority in the room.

Key Takeaways from This Section

1. AI transcription tools (Copilot, Otter.ai, Gemini) let you focus on facilitating during the meeting, not note-taking, which directly improves decision quality.
2. A raw transcript is not useful. The value comes from prompting AI to extract specific artefacts: decision logs, action tables, risk flags.
3. Specific prompts produce specific outputs. Tell the AI the format, the audience, and the purpose, every time.
4. The right tool depends on your existing tech stack. Microsoft Copilot fits M365 users; Google Gemini fits Workspace users; ChatGPT and Claude work for everyone with a paste-and-prompt approach.
5. Always check data and consent rules before recording client or external calls, especially in regulated industries or across different legal jurisdictions.
6. The three-output post-meeting routine (decision log + action table + follow-up email) takes under ten minutes and replaces hours of disorganised catch-up later.
7. Different roles extract different value, from RAID log updates for PMs to structured interview debriefs for HR to discovery call synthesis for sales.

It's Thursday afternoon. You just finished a 90-minute project kickoff meeting with eight stakeholders. Your notes are a mess of half-sentences, three people talked over each other during the risk discussion, and you're staring at a blank document titled 'Meeting Summary. Send by EOD.' You remember the decisions but not who owned them. You remember the concerns but not the exact wording. You have a 4pm deadline and zero energy. This is exactly the situation AI tools were built to rescue.

Turning Raw Meeting Chaos into Structured Outputs

Most professionals use AI for writing emails or drafting reports. Far fewer use it to process meetings, which is a missed opportunity, because meetings are where projects actually live or die. The decisions made in that room, the action items agreed upon, the risks flagged and immediately forgotten, all of that has real project impact. AI can help you capture it, structure it, and distribute it in a fraction of the time it would take manually. The key skill is knowing how to feed your raw notes into an AI tool and ask for exactly the output your team needs.

The process works in two phases. First, you capture, using either an AI meeting assistant like Otter.ai or Microsoft Copilot in Teams, or your own rough notes typed quickly during the call. Second, you process, pasting that raw material into ChatGPT, Claude, or Copilot and asking it to transform the chaos into something usable. Neither phase requires technical skill. The capture phase just requires hitting record or typing faster than you normally would. The processing phase requires knowing how to write a clear, specific instruction to the AI, what's called a prompt.

Think of a prompt like a brief to a very capable assistant who has no context about your company, your project, or your stakeholders. The more specific your brief, the better the output. Telling the AI 'summarize this' produces generic results. Telling it 'extract all action items with owner names and due dates, then list the unresolved decisions that need a follow-up' produces something you can actually paste into your project tracker. That specificity is the entire skill. It takes about two weeks of practice to make it feel natural.

Processing Your Notes Into a Real Meeting Summary

Choosing the Right AI Tool for Meeting Work

Putting It Into Practice on Monday

The fastest way to build this habit is to pick one recurring meeting, your weekly team standup, your client check-in, your project status call, and commit to processing it with AI for four consecutive weeks. Don't try to automate everything at once. Pick one meeting, take your usual notes, then spend five minutes pasting them into ChatGPT or Claude with a structured prompt. Compare the output to what you would have written manually. Most people find the AI version is cleaner, more complete, and takes a tenth of the time.

The second shift is using AI before meetings, not just after. Paste your previous meeting summary into ChatGPT and ask it to generate a focused agenda based on the open items. Ask it to draft the three most important questions you should ask your vendor or client. Ask it to identify which risks from last week's notes still haven't been addressed. This pre-meeting preparation is where project managers find the biggest time savings, because a sharper agenda means a shorter, more productive meeting in the first place.

Over time, you build a repeatable system: AI-assisted agenda before, rough notes during, AI-processed summary after, distributed within 30 minutes of the meeting ending. Stakeholders start noticing. Action items get completed because they're clearly named and dated. Decisions don't get relitigated because there's a written record everyone received. That reliability is what separates project managers who run meetings from project managers who run good projects.

How Different Roles Use AI Meeting Tools

• Project managers: Generate action item logs and risk registers directly from meeting notes, then paste into project trackers like Asana or Monday.com
• Sales managers: summarize client calls into CRM-ready notes with next steps, objections raised, and follow-up timing
• HR professionals: Process interview debrief meetings into structured candidate evaluations with consistent categories across all interviewers
• Teachers and trainers: Turn planning meetings into session outlines with learning objectives and timing breakdowns
• Consultants: Draft client-facing meeting recaps that are polished enough to send without heavy editing
• Executives: Use AI to pre-read long meeting transcripts and surface only the decisions and escalations that require their attention
• Small business owners: Convert supplier or partner meetings into clear agreement summaries they can refer back to without hiring an assistant

Key Takeaways

1. AI tools work best with specific, structured prompts, tell the tool exactly what sections you want in the output
2. Your notes don't need to be polished, fragments with names, decisions, and dates give AI enough to produce clean summaries
3. Use AI before meetings to generate focused agendas from previous action items, cutting meeting time and improving follow-through
4. Different tools suit different setups: Copilot for Microsoft Teams users, Otter.ai for cross-platform recording, ChatGPT or Claude for processing any pasted notes
5. The most powerful habit is a consistent three-step system: AI-assisted agenda before, notes during, AI-processed summary within 30 minutes after
6. Processing meetings with AI creates a written record that reduces decision amnesia, relitigated arguments, and dropped action items
7. Any professional role, from HR to sales to consulting, can apply AI meeting tools to their specific workflow without technical skills