Words I Need To Know

Purpose

Born from the same impulse as the Words backend (vocabulary through push notifications) but evolved far beyond it. Words I Need To Know is a primer app: look up any concept, get a clear definition, then either chat with AI to go deeper or trigger the self-prompting mode where the AI generates expert follow-up questions and answers them, creating a simulated dialogue between specialists that you observe and learn from.

Stack

What I Learned

• The self-prompting feature is the most interesting piece: the AI generates a response to your query, then generates an intelligent follow-up question as if it were a different expert who disagrees or wants to probe deeper, then answers that follow-up, then generates another question, and so on. The result is a simulated Socratic dialogue where you are the audience, not the interrogator. You watch two AI personas explore a topic from multiple angles while you learn from the exchange.
• The technical implementation of self-prompting: after the initial response, inject a system prompt like "You are a skeptical expert in this field. Ask the most important follow-up question that a critical thinker would ask about the previous answer." Capture that question, then switch the system prompt to the original answering persona, generate a response, and repeat. The conversation depth is configurable — 2 rounds for a quick overview, 5 rounds for a deep dive.
• This is functionally a debate simulator — and it produces better learning outcomes than a single authoritative answer because it surfaces the tensions, edge cases, and competing perspectives within a topic. A single answer to "what is inflation?" gives you a definition. A self-prompting dialogue gives you the monetarist view, the Keynesian pushback, the real-world complications, and the areas where economists genuinely disagree. The disagreement IS the learning.
• The primer framing — "Words I Need To Know" — positions the app as a personal dictionary that goes as deep as you want. Surface level: a definition. One layer deeper: a conversational explanation. Deepest: a multi-round expert dialogue. The user controls the depth. Most concepts need a definition. Some need the dialogue. The app serves both without forcing the heavy version on everyone.

Key Insights

• Self-prompting AI is a descendant of the conversations-1 experiment from the AI experiments cluster — simulating adversarial discussion of topics. That experiment used Google Trends for topic selection and produced balanced but toothless output. This version is better because: the user chooses the topic (relevance), the expert personas are tuned per domain (specificity), and the follow-up questions are designed to probe weaknesses in the previous answer (genuine adversarial pressure). The concept matured through iteration.
• Watching experts argue is one of the most effective learning techniques available. Podcast listeners intuitively know this — the best episodes are debates, not monologues. The Words I Need To Know self-prompting feature makes this available on demand for any topic. You do not need to find a podcast where two economists debate inflation. You just type "inflation" and watch the AI simulate the debate.
• The app connects to the 101 / We Are 101 thesis: help people begin. A definition is the beginning. An AI chat is the next step. A self-prompting expert dialogue is the advanced class. The app is a 101 engine — it meets you where you are and takes you as far as you want to go. The We Are 101 LLC failed as a company. The mission survived as a feature.
• For the Potatuhs ecosystem, this app could serve as the educational backbone — "Words I Need To Know about print-on-demand," "Words I Need To Know about game development," "Words I Need To Know about Shopify Hydrogen." Every domain the ecosystem touches becomes a primer topic. The app is a learning interface for the brand's own knowledge.