Coach Tansopalucks drives a lot. During those drives, he brainstorms team strategy, mechanism designs, and notebook structure with AI assistants. Some of those conversations turn into useful insights for the team. Some of them are wrong. AI assistants confidently mix VEX seasons, repeat outdated info, and confuse rule numbers.
This page is where those conversations live after a coach review pass. Every entry shows the original Q&A, what was right, what was wrong, and which of our guides actually answers the question better.
For students reading this page: this is what good engineering review looks like. Senior team members fact-check before junior students consume content, regardless of where the content came from. AI is a tool, not a teacher. Treat what it says the same way you treat a forum post from a stranger: useful, sometimes brilliant, often wrong, always worth verifying.
Topic: passive flip mechanisms, flex wheel sizing for pins/cups, toggle strategy. Coach explored using rubber-band passive flips and flex-wheel intakes for Override scoring objects.
There are real ideas here. There are also real factual errors. Read the bullets below before letting any student use this conversation.
✓ What's solid:
✗ What's wrong:
R11a. The intake counts against the 88W total robot motor cap (R10a). A passive flip uses 0 motor watts — correct — but the "55W limit" framing is the wrong rule.SG10 already prohibits descoring opposing or neutral goals regardless of stacking. The cup doesn't "protect" anything; the rule does. The cup's strategic value is in occupying goal volume, not shielding pins.SC3/SC5). A yellow/yellow pin scores up to 20 pts when both halves are visible and owned. See override-toggle-strategy for the full per-half framework.🎯 What this exposed: we don't have a guide on Override-specific intake geometry. Pin/cup dimensions, flex-wheel sizing, compression vs. rollers vs. claws, and the trade-offs among them. Action: override-intake-geometry guide written, addressing the questions Gemini got wrong.
During the car ride, use the AI assistant's voice mode (Gemini, ChatGPT, Claude). Speak the questions out loud; let the assistant respond. After the conversation, save the transcript — most assistants keep a log automatically and let you export.
Open the transcript. Read it once. Ask yourself:
For each AI response in the transcript, mark one of:
Then write 2–5 bullets of review — what was right, what was wrong, what was the underlying issue (outdated season knowledge, misread rule, etc.).
Some entries expose real gaps in our existing guides. Tag those entries 🎯 Led to new guide and link the new guide. Examples:
Add the entry to the Entries tab in this page. Newest first. Include: