AI Coach & Hint System

Passive reading is penalized at the architectural level. When you log an incorrect answer, the AI Coach activates. Rather than immediately revealing the solution, it presents a structured two-tier intervention designed to force active reasoning.

Tier 1 provides directional guidance without revealing the answer. The prompt is dynamically generated based on the question's topic tag and the specific error type detected in your response. Examples: 'Consider what happens at boundary values when x approaches zero.' Or: 'Check whether the exclusion condition in row 3 is bidirectional or unidirectional.'

Tier 1 hints are designed to activate the reasoning chain you should have used, not to summarize the solution. If you can bridge the gap after a Tier 1 hint, your reasoning is retrievable — it was a momentary block, not a conceptual gap.

Tier 2 shows the complete solution with step-by-step reasoning, including the specific point where the correct approach diverges from your submitted answer. Both tiers are logged to your Telemetry profile. Hint usage counts appear in your session summary and weekly AI Digest.

After every incorrect attempt (with or without hints), the AI Coach tags the error with a Distractor Error Type: Conceptual Gap (the underlying principle was unknown), Execution Error (the principle was known but a calculation or logical step failed), or Trap: Distractor (a specifically engineered wrong answer exploited a predictable reasoning shortcut).

Sub-types include: Trap: Negation (a logically negated version of the correct answer), Root Cause Mismatch (a downstream effect selected instead of the cause), and Calculation Error (sign errors, boundary failures, unit mismatches). These tags are not cosmetic — they are the input to the weekly AI Digest and the SRS trigger logic.

Once per week, the AI Coach generates a personalized digest analyzing your error pattern distribution. The digest surfaces your top two or three error types by frequency, cites the specific question IDs where each pattern appeared, and recommends targeted remediation: Concept Library resources for Conceptual Gaps, slow-repetition practice sets for Execution Errors, and specific trap-awareness drills for Distractor patterns.

Every AI-generated piece of content visible to you — including digest analysis and Socratic prompts — is labeled with an AI-Generated badge. This is a transparency requirement, not optional. You should always know when you are receiving machine-authored feedback.