🤖 Meeting Intelligence Pipeline v2.0

A unified three-stage AI pipeline that chains Email Summarization, Calendar Conflict Resolution, and an enhanced Meeting Agenda Agent — complete with auto-assigned owners, pre-read tasks, and follow-up task lists.

---

✨ What's New in v2.0

Feature	Details
Auto-assign owners	Agenda item owners are automatically assigned from the participant list extracted from your email
Pre-read tasks	AI generates preparation tasks for each participant before the meeting
Follow-up task list	Post-meeting action items are generated with owners and due date suggestions
Email → Agenda chaining	Participants, deadlines, and urgency from your email feed directly into the agenda
Calendar-aware scheduling	Detected conflicts and suggestions flow into the agenda generation context
Unified pipeline	One command runs all three agents end-to-end

---

🏗️ Pipeline Architecture

email.txt          calendar.csv      meeting.txt
     │                   │                │
     ▼                   ▼                │
┌─────────────┐   ┌──────────────┐       │
│   Stage 1   │   │   Stage 2    │       │
│   Email     │   │  Calendar    │       │
│Summarization│   │  Conflict    │       │
│   Agent     │   │  Resolver    │       │
└──────┬──────┘   └──────┬───────┘       │
       │                 │               │
       │  participants   │  conflicts    │
       │  key_points     │  suggestions  │
       │  deadlines      │               │
       └────────┬────────┘               │
                │                        │
                ▼                        ▼
          ┌─────────────────────────────────┐
          │           Stage 3               │
          │   Meeting Agenda Agent          │
          │   (Enhanced)                    │
          │                                 │
          │ • Auto-assign owners            │
          │ • Generate pre-read tasks       │
          │ • Generate follow-up tasks      │
          └─────────────────────────────────┘
                         │
                         ▼
              output/agenda.txt (.json)
              output/email_summary.txt (.json)
              output/conflicts.txt (.json)

---

🚀 Getting Started

Prerequisites

• Python 3.10+
• OpenAI API key

Installation

# Clone or copy this project
cd meeting-intelligence

# Install dependencies
pip install openai pydantic

# Set your API key
export OPENAI_API_KEY="sk-your-key-here"   # macOS/Linux
$env:OPENAI_API_KEY="sk-your-key-here"     # Windows PowerShell

Run the Pipeline

python agent.py

---

📥 Input Files

File	Purpose	Auto-created if missing
email.txt	Raw email text to summarize	✅ Yes (placeholder)
calendar.csv	Scheduled events for conflict detection	✅ Yes (sample data)
meeting.txt	Brief meeting description + duration	✅ Yes (placeholder)

calendar.csv Format

title,start_time,end_time,priority,type,flexible
Team Standup,2025-07-01 09:00,2025-07-01 09:30,high,meeting,no
Client Call,2025-07-01 09:25,2025-07-01 10:00,high,meeting,no

Column	Values
priority	low / medium / high
flexible	yes / no
start_time / end_time	YYYY-MM-DD HH:MM, ISO 8601, or DD/MM/YYYY HH:MM

---

📤 Output Files

All outputs are saved to the output/ directory.

File	Description
output/email_summary.txt	Human-readable email summary with participants, action items, deadlines
output/email_summary.json	Machine-readable email summary
output/conflicts.txt	Human-readable calendar conflict report
output/conflicts.json	Machine-readable conflict data
output/agenda.txt	Full meeting agenda with pre-reads & follow-up tasks
output/agenda.json	Machine-readable agenda (ready for calendar/pipeline integration)

Sample agenda.txt Output

MEETING AGENDA  (2025-07-01)
==================================================
Title    : Q3 Planning Meeting
Objective: Align the team on Q3 roadmap, resolve blockers, assign owners
Duration : 60 minutes
Attendees: Alice, Bob, Carol, Dave

──────────────────────────────────────────────────
PRE-READ TASKS (complete before meeting)
──────────────────────────────────────────────────
1. [Alice] Review Q2 milestone report
   Prepare a 1-page status update covering completed and missed milestones.

2. [Bob] Summarize technical blockers
   Document all open blockers with estimated resolution time.

──────────────────────────────────────────────────
AGENDA
──────────────────────────────────────────────────
1. Q2 Retrospective  (15 min)
   Owner  : Alice
   Outcome: Shared understanding of what shipped and what didn't

2. Q3 Feature Ownership Assignment  (20 min)
   Owner  : Alice
   Outcome: Each feature has a named owner confirmed by the team

3. Design-to-Dev Handoff Process  (15 min)
   Owner  : Carol
   Outcome: Agreed handoff checklist and timeline

4. QA Timeline & Sprint Kickoff  (10 min)
   Owner  : Dave
   Outcome: Sprint start date confirmed and QA plan drafted

──────────────────────────────────────────────────
FOLLOW-UP TASK LIST
──────────────────────────────────────────────────
1. [HIGH] Send feature ownership matrix to all stakeholders
   Owner: Alice  |  Due: within 1 day

2. [HIGH] Begin feature spec documentation
   Owner: Bob  |  Due: by end of next week

3. [MEDIUM] Share design mockup templates with dev team
   Owner: Carol  |  Due: within 2 days

---

⚙️ Configuration

Edit constants at the top of agent.py:

Variable	Default	Description
BUFFER_MINUTES	10	Minimum gap between calendar events
MODEL	gpt-4o-mini	OpenAI model
OUTPUT_DIR	output	Output directory
EMAIL_INPUT_PATH	email.txt	Email input file
CALENDAR_INPUT_PATH	calendar.csv	Calendar input file
MEETING_INPUT_PATH	meeting.txt	Meeting description file

---

🛠️ Tech Stack

Layer	Technology
Language	Python 3.10+
AI Engine	OpenAI GPT-4o-mini
Schema Validation	Pydantic v2
Calendar Parsing	Python Standard Library
API Interface	OpenAI Beta Parsing API (Structured Outputs)

---

📄 License

MIT — free to use, modify, and distribute.