# Ovasabi Studios

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Nobert Momoh](https://cerebralvalley.ai/u/nmxmxh)
- **GitHub:** https://github.com/nmxmxh/ai-and-i
- **Demo video:** https://youtu.be/IPsXWg8vb58
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/30

AI&I: Cognitive Symbiosis Engine

The Problem (Simple Version)

When you use AI, you're getting smarter answers but potentially becoming a lazier thinker.

Think about it: Before Google Maps, you remembered directions. Before calculators, you could do mental math. Before AI, you thought through problems yourself.

AI is different because it doesn't just handle simple tasks - it handles complex thinking. And when you outsource complex thinking, you risk losing the ability to do it yourself.

The pattern:

1. You have a question
2. You ask Claude immediately
3. You accept the answer
4. You never develop your own perspective

Over time, this creates "cognitive atrophy" - your thinking muscles weaken from lack of use.

The Solution (Simple Version)

AI&I is like a gym for your mind when using AI.

The core idea: Before you ask Claude anything, write down what YOU think first.

That's it. That simple act changes everything:

- You discover what you actually know (or don't)
- You notice your hidden assumptions
- You have something to compare against Claude's response
- You can see where you were right, wrong, or limited

The Four Features:

| Feature | What It Does | Why It Matters |
|---------|--------------|----------------|
| Baseline Capture | Records your initial thoughts with a confidence level (0-100%) | Creates a "before" snapshot so you can measure growth |
| Assumption Tracking| Surfaces the beliefs underlying your thinking | Assumptions are invisible until examined |
| Socratic Challenges| AI asks YOU tough questions instead of giving answers | Builds thinking strength, not answer dependency |
| CSI Score | Measures your intellectual autonomy (0-100) | Gamifies cognitive growth over time |

The Four Socratic Challenges

Instead of just answering your questions, Claude can challenge your thinking:

1. Devil's Advocate - Argues against your position, even if it's correct
2. Blind Spot Explorer - Asks what you might be missing or overlooking
3. Feynman Tutor - Tests if you can explain your reasoning simply
4. Logic Auditor - Checks your argument for logical gaps

Example:

- You write: "I think our product is losing to competitors on price"
- Devil's Advocate asks: "What if your competitors are losing money on every sale? What if customers would pay more for better quality?"

This builds the thinking muscle instead of atrophying it.

The CSI Score (Cognitive Symbiosis Index)

A score from 0-100 measuring how well you're maintaining intellectual autonomy:

| Component | What It Measures |
|-----------|------------------|
| Autonomy | Do you think before asking AI? |
| Reflection | Do you reconsider after AI responds? |
| Challenge | Do you seek opposing viewpoints? |
| Divergence | Does your thinking differ from AI's? |

Goal: Not to reject AI, but to use it while staying intellectually strong.

---

Technical Innovation: CWF (Cognitive Wire Format)

The Problem It Solves

Claude has a context window (the amount of text it can "see" at once). Every token you send costs money and uses capacity. JSON is verbose - all those quotes, brackets, and field names add up.

When you have a journal with dozens of entries, session metadata, and assumption tracking, JSON becomes wasteful.

What CWF Does

CWF is a pipe-delimited text format designed specifically for cognitive journal data.

Standard JSON:

```json
{
  "session": {
    "id": "abc-123",
    "title": "Q3 Strategy Review",
    "type": "analysis",
    "baseline_captured": true
  },
  "entries": [
    {
      "type": "baseline",
      "content": "I believe our enterprise position is strong",
      "confidence": 0.75,
      "timestamp": "2024-01-15T10:30:00Z"
    },
    {
      "type": "assumption",
      "content": "SMB needs differ fundamentally from enterprise",
      "status": "unverified",
      "timestamp": "2024-01-15T10:32:00Z"
    }
  ]
}
```

CWF Equivalent:

```
SESSION|Q3 Strategy Review|analysis|true
BASELINE|I believe our enterprise position is strong|0.75
ASSUMPTION|U|SMB needs differ fundamentally from enterprise
```

The Numbers (Verified)

| Metric | JSON | CWF + Brotli | Savings |
|--------|------|--------------|---------|
| Storage Size | 3,195 bytes | 684 bytes | **78%** |
| Token Count | ~180 tokens | ~90 tokens | **50%** |
| Parse Time | ~2ms | ~0.5ms | **75%** |

How we verified: 31 automated tests including:

- Round-trip encoding/decoding
- SHA256 integrity verification
- Random data fuzzing
- Real-world session data

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/30. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt