AI Meeting Preparation Assistant

Google Calendar

Role

Product Manager

Timeline

6 weeks

Tools

Figma, Miro, Google Workspace

01 — Overview

Deliverables

PRD • UX flows • Wireframes • Metrics Framework • Experiment Plan • Product Strategy • Case Study

02 — Context

Google Calendar is the backbone of millions of professionals' workflows. Almost every meeting, event, or collaboration starts here.

But while Google Calendar tells you when you have a meeting… it doesn't help you get ready for one.

Through interviews and behavioral research, I discovered a major productivity gap:

People spend more time preparing for meetings than attending them. And many still show up unprepared.

Across users — engineers, PMs, executives, students — the same problems surfaced:

  • Searching for old emails before a meeting
  • Jumping between docs, threads, and files
  • Forgetting past decisions or action items
  • Missing pre-work shared days before
  • Not having the right context when joining

This case study covers how I designed AI Meeting Prep, a contextual AI assistant built directly into Google Calendar that prepares you for meetings automatically.

03 — Problem

1. Information Lives Everywhere

Important context is scattered across:

Gmail threads
Google Docs
Sheets
Slides
Slack messages
Drive folders
Past calendar notes
Chat history

Users manually dig through all of it before meetings.

2. Meetings Start Without Proper Context

People join calls unsure of:

  • What was discussed last time
  • Who owns what
  • The latest decisions
  • What the meeting is even about

This leads to wasted time and unclear outcomes.

3. Pre-work Is Missed or Ignored

Action items shared before meetings often get buried in email or Slack.

4. Calendars Don't Help You Prepare

Google Calendar alerts users about time, not context. There is no intelligence guiding users toward better preparation.

04 — Opportunity

"How might we transform Google Calendar from a scheduling tool into a proactive meeting assistant?"

An AI assistant inside Calendar could:

Pull related emails
Surface docs and past meeting notes
Summarize the last conversation
Extract action items
Provide recommended agenda topics
Suggest follow-up tasks

This would save users minutes per meeting and improve team alignment.

05 — Hypotheses

If users receive automatic meeting prep summaries, then they will spend less time searching and more time contributing.

If Calendar can extract past decisions and action items, then meetings will start with clear context.

If Calendar shows the most relevant documents and emails, then users will miss fewer pre-work requirements.

If users can generate agendas with one click, then meeting quality and structure will improve.

06 — Personas

Sandra

Engineering Manager

  • • Attends 6–10 meetings per day
  • • Needs quick context before each meeting
  • • Often forgets pre-reads shared through email

Kevin

Product Manager

  • • Joins cross-functional meetings
  • • Juggles docs, threads, and meeting notes
  • • Needs clarity on decisions and action items

Maya

Graduate Student

  • • Uses Calendar for classes, office hours, group work
  • • Needs summaries & doc linking for study sessions

08 — Solution Overview

I designed AI Meeting Prep, an intelligent assistant inside Google Calendar that prepares users for meetings automatically.

It appears directly inside each calendar event.

09 — Core Features

1. Context Summary

AI analyzes:

Previous meeting notesRelated emailsOpen itemsAttached docs

And generates:

Key decisionsDiscussion historyOpen questionsStakeholder responsibilities

2. Task & Action Item Extraction

AI scans across:

EmailsDocsSlides

And extracts:

TasksOwnersDue datesPending follow-ups

Tasks integrate with Google Tasks or Workspace.

3. Relevant Document & Email Surfacing

Using semantic search, Calendar suggests:

  • Docs shared before the meeting
  • Past attachments
  • Slide decks
  • Key email threads

No more digging.

4. One-Click Agenda Builder

Users can generate:

Draft agenda
Talking points
Questions to review
Suggested structure

Editable inside Calendar.

5. Post-Meeting Follow-Up

After the meeting, AI generates a:

  • Summary
  • Task list
  • Decision record
  • Next-steps document

Automatically sent to attendees (optional).

11 — Metrics & Success Criteria

North Star Metric

"Preparation Time Saved per Meeting"

Measured through:

  • • Reduction in time spent switching apps
  • • Time spent reviewing summaries vs searching manually

Supporting KPIs

Feature adoption (% of events using AI Prep)
7-day retention of feature
Document open-through rate
Pre-meeting summary engagement
Task extraction usage
Decrease in meeting duration (optional)
Increase in meeting satisfaction (survey)

Experimentation (A/B Test)

Control

Standard Google Calendar

Variant

AI Prep panel

Measure:

EngagementTask completionPre-read complianceMeeting outcomes

12 — Roadmap

V1 — Core Intelligence

0–3 Months

  • Pre-meeting summary
  • Task extraction
  • Related docs surfacing
  • Draft agenda builder

V2 — Team Workflows

3–6 Months

  • Meeting auto-notes
  • Follow-up tasks
  • Smart reminders
  • Integration into Google Meet

V3 — Autonomous Calendar

6–12 Months

  • Predictive agenda suggestions
  • Meeting duplication detection
  • Calendar optimization ("cancel unnecessary meetings")
  • Cross-tool integrations (Slack, Notion, Asana)

14 — Impact

AI Meeting Prep turns Google Calendar into a proactive partner that:

Saves time
Improves meeting quality
Ensures alignment
Increases productivity
Reduces context-switching
Boosts Workspace retention

Google Calendar becomes more than a scheduling app—it becomes a meeting intelligence hub.

15 — Reflection

This project taught me how to:

  • Connect high-level user pain to workflow automation
  • Understand breadth and depth of AI augmentation opportunities
  • Balance individual vs team needs
  • Focus features around measurable productivity gains
  • Create an AI architecture grounded in real user behavior
  • Blend product strategy with UX simplicity

AI Meeting Prep reflects my approach to building assistant-driven products that work invisibly, intuitively, and contextually—without overwhelming the user.