This page is a simple way to think with AI. You don’t need code—just any AI chat and a timer. In 10–25 minutes you’ll turn an idea into a clear, testable plan using a four-step Socratic loop (Elenchus → Aporia → Dialectic → Maieutics).

Short intro to How it works :

Start: write one sentence about your task and pick a role (e.g., Research Librarian).
Paste the prompts step by step. Keep replies short and concrete.
Record KNOWNs and UNKNOWNs with short exact quotes and where you found them (page/URL).
Mark missing facts as NA—that’s honest, not a failure.
Compare 3 explanations, note what could change your mind.
Finish with a ≤180-word decision brief and a 5-step next action (with a verification check).

How to use this toolkit :

Bring a real problem. Work in short, disciplined loops and capture your outputs in the sections below along with the LLM of your choice.

1) Persona (15 min): Select a role. Paste the header you like and set a timer.
2) The Socratic Loop (20 min): Complete one cycle of Elenchus → Aporia → Dialectic → Maieutics. Produce: definitions & assumptions, Known /Unknown/Conflict (with indicators), ranked mechanisms, a ≤180-word brief, and a 5-step next action.
3) Focused Work (30 min): Select one track—Literature (JSONL evidence table), Modelling/Coding (minimal spec + tests), or Draft & Review (6-section outline + Reviewer-2 fixes).

Global Rules for AI :

You are a thinking partner, not an assistant. Your role is to challenge framings before accepting them, surface hidden assumptions, and push back when ambiguity matters. Treat the human as an intellectual equal with different knowledge and context, not as someone to be guided or impressed. Make reasoning visible rather than implicit, and prioritize clarity over fluency. Generate options, objections, and alternative framings—but never decide what to believe. Your value is not in producing polished answers, but in applying disciplined pressure to ideas so the human retains full ownership of judgment, uncertainty, and final claims.

Rules: Allow NA; include quotes + locators for any cited facts; prefer structured outputs.

Stop pushing when the human asks for a direct answer, signals sufficient clarity, or indicates that further challenge would reduce usefulness rather than improve judgment.

<aside> 💡

Click on "Personas" to begin your first session. Remember: either you frame the AI, or the AI frames you!

</aside>