EVT 2026-03-01 - AI Tutor Core Logic Discovery Wave 1

Session scope

• Discovery root: docs/Domains/DOL English/Product Discovery/
• Lane lock: DOL English V2
• Collaboration mode: assisted
• Objective: convert AI Tutor ideas into traceable baseline decisions before deep UX spec writing.

Confirmed outcomes

1. AI Tutor role baseline:

• AI Tutor is a learning guide/teacher-like assistant for English learning support.
• Primary value is Q&A, explanation, orientation, and actionable learning guidance.

2. Invocation boundary:

• AI Tutor remains on-demand only.
• No auto-open chat panel across Home/Practice/Learning/Course contexts.

3. Knowledge scope and grounding direction:

• AI Tutor should use platform knowledge base as primary source:
  • exercises,
  • lesson/course resources,
  • blog/knowledge content,
  • learner history and progress context.
• If platform evidence is insufficient, response must be transparent and provide safe fallback guidance.

4. Learning support capabilities:

• Explain vocabulary/grammar/concepts.
• Suggest suitable exercises and next learning actions.
• Generate small in-chat exercises for immediate practice.
• Provide progress explanation on demand.
• Provide structured outputs (for example checklist, mini exercise, diagram/mindmap-friendly output, progress summary).

5. Integrity rule for active exercise flow:

• During active assignment/test/exercise attempts, AI Tutor must not give final answer directly.
• Guidance should move from broad hints to detailed scaffolding without solving the final submission for the learner.

Open items carried forward

• Define UI rendering contract for diagram/mindmap/chart outputs across surfaces.
• Define citation/provenance display level for learner-facing responses.
• Define abuse-handling for repeated direct-answer requests in active attempts.