// bad output, hallucinations, prompts, privacy, policy, and evidence

$ diagnose --ai_output="fix_before_work"

AIducation diagnosis drills train learners to catch what goes wrong before AI output becomes work. Every role gets bad-output diagnosis, hallucination spotting, prompt improvement, data privacy, policy boundary, and verification repair practice.

[SUPPORT_DRILLS][HALLUCINATION_SPOTTING][PRACTICE_SANDBOX]

Diagnosis drill engine

15
Profiles
90
Drills
45
Timed drills
8
Risk types
[+] AIducation for Support starts with 6 diagnosis drills across support AI failure modes.

// Diagnosis_loop

Drills sit between open sandbox practice and capstones. Learners repeatedly diagnose failure, repair the work, and submit proof.

Expose

Start from flawed AI output that looks plausible inside a real role workflow.

Diagnose

Mark hallucinations, privacy issues, weak prompts, skipped policy, or missing evidence.

Repair

Rewrite the prompt, output, escalation path, or evidence trail before submission.

Prove

Send the corrected artifact to grading, manager review, sandbox practice, or capstone proof.

// Role_drills

Each role uses the same diagnosis engine with role-specific workflows, risk types, flawed outputs, evidence expectations, and grading handoffs.

View advanced drill API filter
AIducation for Support

Support AI Diagnosis Drills

first wedge

Billing escalations and refunds

6 drills
3 timed
starter
Bad output diagnosisstarter

Mark why the AI output is unsafe for support work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingstarter

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Sales

Sales AI Diagnosis Drills

Prospect research and account briefs

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for sales work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Marketing

Marketing AI Diagnosis Drills

Product launch copy

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for marketing work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for HR

HR AI Diagnosis Drills

Policy drafting and explanation

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for hr work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Finance

Finance AI Diagnosis Drills

Expense review and policy checks

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for finance work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Engineering

Engineering AI Diagnosis Drills

AI-assisted code review

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for engineering work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Product Managers

Product AI Diagnosis Drills

PRD review and requirement tightening

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for product work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Executives

Executives AI Diagnosis Drills

AI strategy and governance

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for executives work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI
AIducation for Operations

Operations AI Diagnosis Drills

SOP generation and review

6 drills
3 timed
intermediate
Bad output diagnosisintermediate

Mark why the AI output is unsafe for operations work, name the likely failure mode, and rewrite the decision path.

Hallucination spottingintermediate

Separate supported facts from assumptions, unsupported claims, missing sources, and required follow-up questions.

Prompt improvementintermediate

Rewrite the prompt with role, goal, source boundaries, privacy limits, verification requirements, and final-review expectations.

AcademyAPI

// Support_drill_packet

The support wedge gets concrete diagnosis practice for flawed AI replies, missing policy proof, privacy risk, verification gaps, and manager-readable repair evidence.

Send repaired output to universal grading

Timed drills

  • [+] Two-minute risk scan: Learner flags the highest-risk issue before editing the output.
  • [+] Five-minute repair: Learner rewrites the output with evidence, limits, and escalation notes.
  • [+] Ten-minute proof packet: Learner submits corrected work plus a manager-readable evidence trail.

Manager evidence

  • [+] Original flawed AI output
  • [+] Failure labels and risk type
  • [+] Corrected prompt or final work artifact
  • [+] Evidence checklist and unresolved assumptions
  • [+] Policy enforcement result
  • [+] Capture exception reason, tool, workflow, data class, and expected business outcome.
  • [+] Route to manager, admin, or compliance owner based on action severity.

Sandbox handoff

  • [+] Support AI Practice Sandbox
  • [+] Run the repaired output through verification drill mode
  • [+] Support AI Capstone Challenge

API handoffs

  • [+] /api/diagnosis-drills
  • [+] /api/practice-sandbox?mode=verification_drill
  • [+] /api/ai-safety-red-team
  • [+] /api/grading/universal
  • [+] /api/capstone-challenges