×

Asking Better Questions

Most people blame their answers — The problem was always the question.

You already know how to use AI. You've tried the prompts. You've watched the tutorials. You've gotten useful outputs and forgettable ones and you're not always sure why.

Here's what nobody tells you: the tool isn't the variable. You are.

The quality of what AI gives you is a direct reflection of the quality of your thinking going in. Which means the people getting extraordinary results aren't better at AI. They're better at asking.

This course teaches that. Four Sunday afternoons. Small room. Real problems. No theory that doesn't immediately touch your actual life.

×

What is the Diet Coke Oracle System?

This system is not exactly just a magic 8-ball, and it cannot predict the future. It is a structural refinery that borrows ancient logic of the I Ching to force a change in your perspective.

The Oracle works by making you strip away the emotional noise from your problems so you can look at them as pure, mechanical systems.

Privacy / How it was made

This site does not gather or store any of your personal information; it is connected to a pretrained Gemma4 LLM model being run locally on Jesse's computer, routed through a Cloudflare worker, so that this site can run as a flat "application". It does not retain any information on you or your conversations; the history archive feature only stores information locally within your browser cache (if your browser is compatible). Jesse does keep the input/output history as it is the data this site sends when you click send on a message, but other than what is typed into the chat bar (through keyboard or speech to text) is kept local on your machine. All data is encrypted in transit.

Diet Coke is built using a quite large 31B parameter opensource LLM (Gemma4), that has been pretrained using LoRA on ~70000 initial question / answer pairs (synthetically derived from ~700 real hand written pairs) and prompting templates borrowing from both The Church of Conceptual Art, and a number of our favorite English translations / versions of the I Ching.

How the Phases Work

Mode 1: The Refinery
The AI acts as a strict structural editor to refine what your real conflict is. It is explicitly trained to reject vague complaints, emotional venting, or purely predictive fortune-telling. It demands that you reframe your problem into a precise, structural logic-gate before it will allow you to proceed.

Mode 2: The Casting
Once you achieve structural clarity, you cast the lines. You will answer a series of fixed binary questions about the tension of your situation (e.g., Assert vs. Submit, Internal vs. External). Your choices, combined with probability, generate a specific hexagram.

Mode 3: The Interpretation
The Oracle maps the Wilhelm translation of the I Ching directly onto the clinical, refined parameters you established in Phase 1, at the time of the output.

Mode 4: The Discussion
You have completed the casting. You can engage in a brief, clinical, structural discussion about the result after the oracle has given you your Hexagram readings.

How to Get the Best Results

To get past the Refinery and get a highly accurate interpretation, you must play by the Oracle's rules:

• Bring a specific target: Do not ask broadly about "life." Bring a specific contract, a specific argument, or a specific creative bottleneck.
• Questions about insight are valid: If you seek internal clarity rather than predictive answers (e.g., "What insight will I gain by making X decision?"), the Oracle will accept it.
• Be clinical: Remove your feelings. Describe the situation using structural terms like "friction," "leverage," "architecture," or "metrics."
• Accept the pushback: The AI's rejections are part of the process. Use its prompts to narrow down exactly what is functionally broken or misaligned in your situation.

×

Archive

Your local consultation history.

CLEAR ARCHIVE