Build Cora, your AI Team Lead - a system that learns from conversations and evolves its own playbook.
Processes your meeting transcripts, maintains context in Notion, creates tasks automatically, and spawns specialized sub-agents for complex work.
As your AI Team Lead, Cora orchestrates your AI workforce—reading conversations, updating knowledge, and delegating tasks to specialized sub-agents. Notion stores her persistent playbook with your preferences, patterns, and rules. Her sub-agents handle specific jobs like research or meeting prep, then disappear.
graph LR
A[You Speak] --> B[Cora Listens]
B --> C[Updates Playbook]
C --> D[Manages AI Team]
D --> E[Team Executes]
E --> F[Learns & Improves]
F --> B
style B fill:#e1f5ff
style C fill:#d4edda
style E fill:#fff3cd
Core capabilities:
- Processes conversations via Limitless or Fireflies
- Maintains preferences and context in Notion
- Creates tasks, prepares meetings, manages projects
- Updates its own playbook from your feedback
- Spawns specialized sub-agents for complex work
graph TB
subgraph "Your World"
Y[Your Conversations]
T[Your Tasks]
F[Your Files]
end
subgraph "Cora's AI Team"
CC[Cora - AI Team Lead]
S1[Research Agent]
S2[Task Manager]
S3[Meeting Prep]
SN[New Agents...]
end
subgraph "Living Memory"
N[Notion Playbook<br/>Rules, Knowledge, Preferences]
end
Y --> CC
T <--> CC
F --> CC
CC <--> N
CC --> S1
CC --> S2
CC --> S3
CC --> SN
style CC fill:#e1f5ff
style N fill:#d4edda
Each sub-agent is a fresh Claude Code process with a focused purpose, scoped access to only the tools it needs, and a clean lifecycle: created → executes → returns → destroyed.
Example voice command:
"Cora, find dinner reservations for 4 people tomorrow"
Behind the scenes:
# Cora spawns a specialized sub-agent
claude-code --system-prompt "Extract dinner preferences and find restaurants" \
--context-files limitless-transcripts.json \
--tools opentable,mapsDinner Planning
Scans your calendar, extracts restaurant preferences from past conversations, proposes three options with rationale.
Meeting Prep
Compiles context from related conversations, previous meetings, and relevant documents 30 minutes before each meeting.
Task Triage
Reviews incoming tasks, assigns priorities based on your patterns, flags items needing immediate attention.
graph LR
A[📋 1x: Organize<br/>Clean Data] --> B[🤖 10x: Automate<br/>Claude Code] --> C[🎨 100x: Create<br/>Move Fast]
style A fill:#fff3cd
style B fill:#d1ecf1
style C fill:#d4edda
- 1x: Get organized - Clean knowledge base in Notion
- 10x: Deploy Cora - Your AI Team Lead manages execution
- 100x: Pure creativity - You focus on vision while your AI team handles operations
Read the full 100x Framework →
Core:
- Claude Code: Universal runtime and orchestration
- Notion: Knowledge base and playbook storage
- MCP Servers: Tool integrations
Integrations:
- Limitless/Fireflies: Conversation capture
- ClickUp: Task management
- Dropbox: File access
Infrastructure:
- No databases to manage
- No custom code to maintain
- No deployment pipelines
- Just Claude Code + APIs
Traditional AI assistants:
- Fixed capabilities that don't grow
- Complex infrastructure to maintain
- Rigid workflows that break easily
- No memory between sessions
This implementation:
- Builds new capabilities through sub-agents
- No infrastructure—just APIs
- Adapts patterns based on feedback
- Notion provides persistent memory
- Claude Code Architecture - Technical deep dive
- 100x Framework - Vision and principles
- Claude Code is the runtime: Not a tool, the execution environment
- Notion is the knowledge base: All memory and learning persists there
- Sub-agents are ephemeral: Create, execute, destroy keeps it clean
- Conversations are training data: Every correction becomes a rule
- Start simple, grow complex: Basic loops first, sophistication emerges
Fair use for personal productivity. See LICENSE for details.
AI that amplifies human capability.
From chaos to clarity. From ideas to execution. From 1x to 100x.
