Martin - The Vision
Executive Summary
- Purpose: Martin - The Vision
- Key sections: Three Core Principles, Design Principles
- Focus: First principles design: augmentation, intelligence, and learning
Three Core Principles
Augmentation, Not Automation
Don't automate CSM tasks - provide data-driven recommendations. Don't remove CSMs from the loop - give CSMs superpowers.
Intelligence, Not Just Speed
Don't just pull data faster - synthesize insights across sources. Don't just answer questions - learn from execution and improve.
Learning, Not Just Execution
Don't just run workflows - detect patterns and trends. Don't just report status - predict future states.
Design Principles
Portfolio-First Architecture
Every CSM conversation started with "load my portfolio" - a manual administrative step repeated dozens of times per week.
- Automatic, silent portfolio load on first message (Phase -1 Bootstrap)
- Portfolio context available for all subsequent queries
- Significant time saved per session (many sessions/week)
Mode-Driven Execution
Users wanted both strategic advice (what should I do...) and immediate execution (do it now). A single interface couldn't serve both needs.
- /plan - Analyze, recommend, explain (no logging)
- /act - Execute play, generate artifact, log results
- Flexibility without confusion
Persistent Intelligence
Early versions had no memory across conversations. Every execution started from scratch - no longitudinal tracking, no state change detection.
- Scalable Snowflake architecture
- Execution archive, account state, risk history, runner logs
- State change detection (new risks, escalations, improvements)
Conversational Interface
CSMs don't want to learn a new tool. They want to ask questions in plain English and get actionable answers.
- Natural language -> structured action
- Skills Registry maps commands to Play DNA
- Fuzzy matching for account targeting