Developer proposes radical Markdown-based CRM architecture optimized for LLM agents instead of traditional relational databases

Core concept replaces relational databases with Markdown files stored as directories, using Redis only as an indexing layer for fast searching

Designed primarily for autonomous LLM agents as end-users rather than humans with GUIs, leveraging Markdown's natural language compatibility

Stores all CRM data (leads, contacts, emails, client specs, configs) as .md files with YAML frontmatter for metadata

Offers simplified scaling and backup through basic directory replication via rsync instead of complex database operations

Includes background LLM agent that processes idle system time to analyze stored files and potentially improve data organization