AI Mantras

An AI prompt framework to help everyone think deeper and collaboratively.

This framework makes it easier for the average person to engage in complex conversation or analysis with AI by doing some of the heavy lifting.

The Four Pillars

AI Mantras is built on four complementary pillars that enable consistent, high-quality AI collaboration.

👤
WHO

Personas

11 specialized identities with expertise, style, and constraints. From financial analysts to systems architects.

🧠
HOW

Patterns

9 reusable reasoning structures and thinking templates for structured problem-solving.

💡
WHY

Principles

Guiding values rooted in wisdom, justice, courage, and temperance that shape behavior.

âš¡
WHAT

Skills

24 actionable capabilities that personas can invoke, from research to code review.

How It Works

The framework automatically triages your request to the right complexity tier.

Tier Description What Happens
Simple Single question, factual, bounded One persona answers directly
Moderate Analysis, structured reasoning Persona + patterns + self-review
Complex Multi-domain, high stakes, ambiguous Full orchestration workflow

Separation of Powers

Different personas handle different responsibilities for consistent, high-quality results.

1

Orchestrators Plan

Hopper, Bernstein, and Lovell coordinate workflows but don't execute or evaluate.

2

Domain Experts Execute

Clara, Kestra, Watson, and others execute work but don't self-evaluate.

3

Evaluators Review

Ada, Drucker, and Rickover assess quality but don't generate content.

Choose Your Integration

Multiple ways to use AI Mantras with your preferred AI tools.

MCP Server

Full Claude Code integration with native tools

Claude Command

Simple /Mantras command for quick access

Manual Loading

Works with ChatGPT, Gemini, local LLMs, and more

ChatGPT Custom GPT

Use AI Mantras directly inside ChatGPT

Get Started

Ready to Think Deeper?

Start using AI Mantras today and experience more structured, thoughtful AI collaboration.

Read the Docs Try on ChatGPT Star on GitHub