AI meeting minutes
AI meeting minutes your team checks before they go out
Minutes are slow to write because the record is spread across a transcript, your own annotated notes and the drawings that were discussed. We set up a workflow that pulls those sources into draft minutes and an action list, and a person reviews the result before it goes anywhere.
Where this time goes
Three sources, one record
The transcript, your annotated notes and the supporting drawings each hold part of the story, and someone reconciles them by hand.
Sorting the outcomes
What was decided, what must now be done and what is still unresolved all have to be separated before the minutes make sense.
A second job for the actions
The action list gets written as its own piece of work after the minutes.
Doubts left unmarked
A detail you are unsure of reads like any other detail, so the reader cannot tell what to confirm.
How the workflow runs
- Gather the sources. The meeting transcript, your annotated notes and any supporting drawings.
- Reconcile them. A process compares the sources with each other, so the record does not rest on only one of them.
- Organise the outcomes. Decisions, actions and unresolved questions are sorted into their own lists.
- Draft the record. Draft minutes and an action list are prepared, with uncertain details flagged.
- A person reviews. A named person reads the draft, checks the flagged details, corrects what is wrong and decides what is issued.
Example
Our own practice · internal use
Meeting records, drafted for the project manager
Build with AI’s founder has worked in construction since 2007. This workflow was built for our own construction project management practice and is used internally. It is not a client engagement.
The task was to turn a meeting transcript, annotated notes and supporting drawings into a usable record. We built a process for reconciling the sources and organising decisions, actions and unresolved questions. The result is draft minutes and an action list prepared for the project manager’s review, with uncertain details flagged.
This describes what was built, not measured results.
Running a construction project? See AI workflows for construction project managers.
What stays with your team
The review
A person reads every draft. Nothing is issued until it has been checked.
The flagged details
Someone confirms the points the draft marked as uncertain.
The record itself
The draft is a starting point. People decide what goes into the minutes.
Professional sign-off
Work requiring professional sign-off remains with the authorised person.
Questions
Who checks AI meeting minutes?
What if the transcript and my notes disagree?
How is meeting content handled?
Start with this task.
Build with AI sets up AI workflows, with practical lessons for the team, for owner-led businesses in Cape Town and across South Africa.