AI Executive Brief
BUILD · MEETING PREPARATION
BUILD · MEETING PREPARATION
This page shows an AI meeting-prep workflow I built with Google Apps Script and the Gemini API. You’ll see the system map, the source log where each item is classified, and the finished executive brief it produces.
Before a meeting, the important details are spread across emails, calendar events, documents, and spreadsheets. Reviewing each one separately takes time, and it’s easy to miss a decision or deadline. The workflow pulls from four sources (Google Calendar, labeled Gmail, a Google Doc, and a Google Sheet), uses Gemini to sort what it finds into decisions, risks, deadlines, and follow-ups, and writes a brief in Google Docs with every point traced back to its source. I tested it with sample meeting data, and a second test brief ran with no code changes.
I wanted to spend less time gathering information before meetings and more time deciding. Every brief ends with a note that a person reviews it before anyone relies on it, because AI does the first pass and a person makes the call.
Executive updates often arrive in different places and formats, including email, documents, and spreadsheets.
When leadership has to read through each source separately before a meeting, preparation becomes repetitive, and important decisions, blockers, or follow-up items can be missed.
The business need was clear: one view of what required a decision, ready before the meeting started.
I designed a three-stage pipeline (Pull, Classify, Generate) controlled from a Google Sheet, with every source logged before it’s classified.
The workflow pulls out what requires a decision, identifies risks and deadlines, and highlights follow-up actions, so the most important information is in one place. Every point traces back to where it came from, so the brief can be validated against its sources before anyone relies on it.
1. The meeting is queued
A meeting is added to the Briefing Queue and marked Ready. I then run the stages in order.
2. Updates are collected
Relevant information is gathered from four sources: the calendar event, labeled emails, a working document, and a metrics spreadsheet.
3. Every source is logged
Each source is recorded in one log and tagged to the meeting, so nothing is lost or mixed up with another meeting.
4. Information is classified
AI reads each source and classifies it against four defined categories: decisions needed, risks, deadlines, and follow-up.
5. The executive brief is generated
The results are organized into one consistent briefing document, with every point traced back to its source and a note that a person reviews it before anyone acts.
6. Errors are logged
Pull and generate errors are written to the control sheet. If AI classification fails, it retries up to three times, and the failure is logged.
Google Apps Script · Gemini API · Gmail · Google Calendar · Google Sheets · Google Docs · ChatGPT · Claude
The workflow brings updates from four sources into one brief, so meeting preparation starts with review instead of searching. On the BRIEF-001 test, a calendar event, labeled emails, a working document, and a metrics sheet produced a single brief. I then refactored the pipeline to run on any brief marked Ready in the queue, and a second test brief ran with no code changes.
It also creates a more consistent meeting-preparation process by surfacing decisions, risks, deadlines, and follow-up in the same format every time, with each point traced to its source.
The shift is from searching and reading to reviewing and deciding. AI does the first pass; a person makes the call.
Pull · Classify · Generate
Shows how the four sources flow through collection, AI classification, and brief generation, controlled from one sheet. (Select to enlarge.)
Source Log
Shows each source logged for BRIEF-001, with decisions, risks, deadlines, and follow-ups extracted by Gemini. (Select to enlarge.)
Executive Briefing Output
Shows the brief generated from BRIEF-001, organized by section with each point traced to its source. (Select to enlarge.)