MEMO (Meeting Extraction, Monitoring & Orchestration) is a multi-agent intelligence system that transforms raw, unstructured meeting transcripts into structured, actionable intelligence. Instead of manually reviewing meeting notes, MEMO automatically identifies who said what, what decisions were made, what action items were assigned, and which items need immediate escalation.
⚡ How It Works
You paste or upload a meeting transcript. MEMO processes it through a pipeline of 6 autonomous agents that run sequentially, each building on the previous agent's output:
Step 1 — Extraction
🔍 TranscriptAnalyzer
Parses the raw transcript to identify speakers, topics, and conversation segments. Structures the unformatted text into analyzable sections.
Step 2 — Extraction
📋 DecisionExtractor
Identifies explicit and implicit decisions made during the meeting. Captures decision context, rationale, and who was involved in making each decision.
Step 3 — Extraction
✅ ActionItemExtractor
Pulls out all action items, to-dos, and follow-ups mentioned during the meeting. Captures deadlines, responsible parties, and dependencies.
Step 4 — Decision
📊 TaskPrioritizer
Evaluates each extracted action item and assigns a priority level (Critical, High, Medium, Low) based on urgency, impact, and deadlines.
Step 5 — Decision
👤 OwnerAssigner
Determines who should own each action item based on context clues, mentions, and responsibility patterns identified in the transcript.
Step 6 — Decision
🚨 EscalationDecider
Evaluates whether any items require immediate escalation — blocked tasks, missed deadlines, cross-team dependencies, or unresolved conflicts.
🎬 Live Agent Pipeline Demo
Watch how MEMO processes a meeting transcript through all 6 agents in real-time
Alice: Let's discuss the Q3 roadmap. The API deadline is Friday. Bob: I'll handle the backend integration. Need auth tokens from DevOps. Alice: Agreed. Bob, please have the API docs ready by Thursday. Charlie: I'm blocked on the frontend — waiting for Bob's endpoints.
🔎
Step 1
TranscriptAnalyzer
Idle
3 speakers · 4 segments parsed
📋
Step 2
DecisionExtractor
Idle
2 decisions: API priority + deadline
✅
Step 3
ActionItemExtractor
Idle
3 actions: API, docs, auth tokens
📈
Step 4
TaskPrioritizer
Idle
P1: API · P2: Docs, Tokens
👤
Step 5
OwnerAssigner
Idle
Bob → API, Docs · DevOps → Tokens
🚨
Step 6
EscalationDecider
Idle
⚠ Charlie blocked → Escalate
✅ Pipeline Complete — 6 agents processed in 4.2s
2
Decisions
3
Action Items
3
Assigned
1
Escalations
🎯 Real-World Use Cases
Engineering Standups
Automatically extract blockers, decisions, and assigned tasks from daily standup transcripts — no more forgotten action items.
Board Meeting Minutes
Turn lengthy board meeting recordings into structured minutes with clear decisions, action items, and accountability assignments.
Client Call Tracking
Capture client requirements, commitments, and follow-up tasks from sales or customer success calls to ensure nothing falls through the cracks.
Sprint Planning & Retrospectives
Extract sprint goals, assigned stories, retrospective action items, and improvement commitments from agile ceremonies.
🏗️ Architecture
🔄
Orchestration Engine
Coordinates all 6 agents in sequence, managing state and message passing between each step.
🛡️
Circuit Breaker Recovery
Each agent has its own circuit breaker. If failures spike, the circuit opens to prevent cascading failures.
📋
Full Audit Trail
Every decision, confidence score, and state change is logged immutably for compliance and traceability.
📤 Upload Your Own Transcript LIVE AGENTS
📄
Drag & drop a file here, or click to browse
.txt · .md · .log supported
✦ Your transcript is sent to the live Python agent pipeline — real extraction, real decisions, auditable results.
• Task requires API integration skills
• Bob Wilson has matching skills: api, integration
• Bob Wilson has lowest current workload (1 task)
• Selected with confidence 0.85
Confidence: 85%
⚠ priority_assigned — Security Audit P1
• High business impact: security keyword detected
• Deadline within 3 days of current date
• Business priority match: "security"
• Scored 72/100 → P1 High
Confidence: 85%
📋 decision_extracted — Launch Date
• Keyword "decided to" detected → explicit decision
• Speaker: Sarah Chen (role: manager)
• Full agreement: all participants present
• New deadline: January 10th