First Principle AI

Inspiration

The inspiration for First Principle AI came from a problem we kept running into while using AI chatbots for learning. When we ask ChatGPT, Claude, or Gemini to explain a difficult topic, the answer may start out helpful, but the conversation quickly becomes long, messy, and hard to follow.

If we get confused halfway through and ask a follow-up question, the chat continues downward. After a few more questions, we lose track of the original explanation, where we stopped, what we already understood, and what concept we were trying to learn.

This is especially painful for STEM topics, projects, research, and complex systems, where one idea depends on many smaller ideas. We wanted to redesign the AI learning experience so users can understand any subject from first principles without losing context.

Instead of a linear chatbot, we imagined a recursive block-based learning interface where users can start with a high-level explanation, dive deeper into confusing concepts, and then return back to the original topic.

What it does

First Principle AI is an AI learning platform that explains any subject through main blocks, subblocks, and recursive learning paths.

The user starts by setting up a profile with their background, knowledge level, learning style, goals, and preferences. Then they can ask a question like they would in a normal AI chat interface.

For example:

How do earbuds work?

The AI first asks follow-up questions about the user’s background, objective, desired depth, and constraints. Then it creates a Main Block, which acts like a personalized mini textbook chapter.

If the user does not understand something inside the Main Block, they can highlight any word, sentence, equation, image, or concept and create a Subblock. That Subblock appears to the right of the parent block and explains that specific concept in more depth.

This can continue recursively:

How earbuds work
→ Speaker driver
→ Electromagnet
→ Magnetic field
→ Moving electric charge
→ Electron

Built With

• amazon-web-services
• cloudfont
• clouldflare
• fastapi
• openai
• python
• react
• s3