To avoid generic LLM responses, a user "trains" her agents by providing them with an identity built on literature. By telling an agent it has read and finds specific books fascinating, its outputs become quirkier and more aligned with a desired persona.

Instead of one generalist AI assistant, create multiple specialized agents, each with a unique persona (e.g., a creative teacher) defined in a "soul" file. Partition their access to specific data "vaults" (like separate Obsidian folders). This specialization improves output quality and maintains logical, secure boundaries between different life domains.

Instead of a single, monolithic "About Me" file, structure personal context into modular files (e.g., roles, projects, team). This design allows you to provide an AI agent with only the specific information it needs for a given task, which enhances efficiency, relevance, and privacy.

Managing permissions for AI agents is a huge challenge. The most likely near-term solution is not granular, per-app controls, which create overwhelming cognitive load. Instead, agent identity will be managed through distinct user personas, like a "work agent" for professional tasks and a "home agent" for personal ones.

Instead of treating the AI as a faceless tool, assign it a full name (e.g., "Zane Calder"). Use this name to create its dedicated Mac user account, email address, and other logins. This reinforces the concept of a separate, autonomous digital assistant.

The 'Claudie' AI project manager reads a core markdown file every time it runs, which acts as a permanent job description. This file defines its role, key principles, and context. This provides the agent with a stable identity, similar to a human employee, ensuring consistent and reliable work.

SaaStr avoids a single, monolithic AI. Instead, they create distinct agents (VP of Marketing, VP of Customer Success) and treat them as separate entities. This architectural choice keeps them focused and allows for tailored interactions without creating a complex, all-knowing system.

To create a highly personalized agent, don't just write its personality file. Instead, ask the new agent to generate a questionnaire about your goals, then answer its questions to give it deep, specific context for its own setup.

Instead of a monolithic AI, create a team of agents with specific roles (e.g., 'Debbie the assistant,' 'Soren the engineer'). This human-like model makes it easier to manage capabilities, control access, and conceptualize the system's functions because it maps to our innate understanding of human teams.