project managementai tools

How I Use AI in Technical Meetings as a Project Manager

The most useful thing AI has done for me as a project manager wasn't writing an email. It was explaining a meeting I'd just sat through.

Here's the short version: I record technical meetings in Teams, ask Copilot to restate the transcript in plain language and list the decisions, then check its summary against the recording before I send anything. That's how I use AI in meetings as a project manager, and it's changed how I run technical projects.

Here's the meeting that made it click.

What was the meeting about?

It was a working session to change how pricing data is stored in the company's master data management system (MDM).

My project is about enabling the company to provision new products. To do that, we had to adjust how pricing information lives in the MDM. That meant getting into the details: how the data is formatted, which tables and columns are involved, and what had to change.

The room had everyone you'd expect:

  • Pricing experts
  • The data team that manages the MDM profile
  • Solutions architects
  • Engineers

Within ten minutes, the conversation was all table names, field names and formatting rules.

My job wasn't to design the change. It was to make sure everyone left agreeing on it, and then explain it to people who weren't technical or weren't there. Honestly? Following every detail in real time while also running the meeting is hard.

How did I use Copilot to make sense of it?

I recorded the meeting in Teams, then asked Copilot to translate the transcript for a less technical audience.

After the call, I opened the transcript and asked Copilot to do two things:

  1. Simplify the technical language so someone outside the data team could follow it.
  2. Outline the outcomes and decisions we'd made.

If you want to try it, a prompt along these lines works:

Here's the transcript of a technical meeting about changing how pricing data is stored in our MDM. Explain the outcome in plain language for someone who isn't technical. List each decision we made, what's still open, and who owns the next steps.

What came back was a version of the meeting I could actually explain. It skipped the word-for-word detail and gave me the "so what" of each decision.

Why did I still watch the recording?

Because transcripts get technical names wrong, and in a data meeting the names are the whole point.

When people say a column or field name out loud, the transcript doesn't always capture it exactly. A table name gets split into two words. A field name comes out as something that sounds similar but doesn't exist.

This is where the Teams recording earned its place. People had been sharing their screens, so I could pause the video and look at the actual table. The real column names. The real field names. Then I checked that what Copilot said matched what was on screen.

AI gave me the first draft. The recording let me verify it.

I wouldn't send a technical summary without that step. If a field name is wrong in the recap, someone downstream builds on the wrong thing.

What about the people who missed it?

They got three ways to catch up, and only one of them needed me.

Not everyone who needed to know about the decisions could attend. Afterwards they had:

  • My email, with the plain-language summary and decisions
  • The recording, so they could watch the discussion and see the screen shares themselves
  • The transcript, which they could ask Copilot about directly, like "what did we decide about the price format?"

That last one surprised me. People didn't need me to re-explain the meeting. They could ask their own questions and get answers based on what was actually said.

What would I tell another project manager trying this?

Start with the tool your company already has, and treat the AI summary as a draft you check.

What I've learned so far:

  • Say you're recording. Tell people at the start, and follow your company's policy on recording and AI tools.
  • Say decisions out loud. "So we're agreed: the price field changes to this format." Clear statements make clear transcripts.
  • Ask for two versions. One so you understand it, and one for the wider group.
  • Check names against the screen share. Tables, columns, systems and acronyms are where transcripts slip.
  • Ask for action items with owners, and allow "unclear". If the owner wasn't stated, you want to know, not have AI guess.
  • Share the recording with the summary. It lets people check things for themselves.

The skill isn't knowing every technical detail. It's making sure everyone leaves with the same understanding.

Which AI meeting tools can project managers use?

I used Teams with Copilot because that's what the company runs on, but most meeting platforms now have something similar.

If you're not in a Microsoft environment, or you want something that works across platforms, these are worth a look:

ToolWhat it does for meetings
Microsoft Teams + CopilotWhat I used. Recording and transcription with Microsoft 365. Intelligent Recap adds AI notes and suggested action items, and Copilot answers questions about the meeting. Recap needs Teams Premium or a Microsoft 365 Copilot license.
Zoom AI CompanionMeeting summaries, next steps, and catch-up questions if you join late. Included with paid Zoom plans.
Google Meet with Gemini"Take notes for me" writes notes into a Google Doc.
GranolaAn AI notepad that doesn't join the call as a bot. You jot rough notes, and it fills them out from the audio.
Superwhisper MeetingsBot-free recording on Mac, with transcription and summaries done on your device and speaker labels.
Wispr Flow NotetakerBot-free, with live transcript, summaries, action items and searchable history. Records on your device, so nothing joins the call.
Otter.ai / Fireflies.aiJoin the call as a visible bot. Live transcript, summaries and action items.

One thing these don't all give you is the video. In a technical meeting, being able to see the screen share was what let me trust the summary.

Have you used AI to make sense of a technical meeting? I'm curious what worked for you, and what it got wrong.

FAQ

Yes, as a starting point. Give it the transcript and ask for the outcome in plain language, the decisions made, and what's still open. Then check any technical names against the recording or screen share, because transcripts often get table, column and system names wrong.

Record and transcribe the meeting in Teams. Afterwards, open Copilot on the meeting and ask it questions about the transcript, such as a plain-language summary or a list of decisions. Some recap and Copilot features need a Teams Premium or Microsoft 365 Copilot license, so check what your company has.

In technical meetings, the details people say out loud, like table names, column names and field names, are often transcribed slightly wrong. If people shared their screens, the video shows the real names. Checking takes a few minutes and stops errors from reaching the people who build on your summary.

Send them your plain-language summary along with the recording and transcript. With Copilot in Teams, they can ask the transcript their own questions, like what was decided about a specific item, without waiting for someone to re-explain it.

Often not. Teams, Zoom and Google Meet all have built-in AI meeting features on many business plans, though some need an add-on license. Standalone tools like Granola, Superwhisper or Wispr Flow are worth a look if your platform doesn't offer what you need.