Guided Mini-Walkthrough — Your AI Workflow

Setup

This is your full rehearsal. You're going to run through the complete AI-assisted work cycle on a simpler task — the same process you'll use in the capstone, but with lower stakes. By the end, you'll have a "mini Real Tuesday" with all 6 sections completed.

The Complete AI-Assisted Work Cycle

This is the workflow you've been building piece by piece. Now you'll run it end to end.

Your Turn

Step 1: Pick your task. Choose one:

• A real work task you do frequently (different from the one you'll use for the capstone)
• Or select from these: (a) write a meeting recap email from bullet-point notes, (b) summarize the key points from a report or article you recently read, (c) draft a brief for a colleague explaining a decision or process

Step 2: Assess the task. Place it on the reliability spectrum. Name: output type, verification method, failure consequence.

Step 3: Classify data sensitivity BEFORE submission. List every piece of information you plan to share with the AI. Classify by sensitivity level (Public, Internal, Confidential, Restricted) and pick the appropriate service tier. If you have Confidential or Restricted data, decide your mitigation NOW: anonymize, omit, or use a paid/enterprise tool. This is a pre-submission gate, not a retrospective check.

Step 4: Select your tool category. General chatbot, research engine, document-grounded, or specialist? Why? Verify your selection works with the data tier from Step 3.

Step 5: Construct your four-element prompt. Role, context, format, constraints — all labeled. Include only the information you cleared in Step 3.

Step 6: Submit and save the complete, unedited output.

Step 7: Diagnose. Find 3+ failures with quoted text, named category, and consequence.

Step 8: Evaluate against professional standards. Name 2+ standards relevant to this task. Find 3+ revisions with original text, revised text, standard cited, and rationale.

Step 9: Revise and produce your final version. Apply all revisions. Save the final artifact alongside the original AI output.

Quality Check

Your mini Real Tuesday should have all of these (same 9-box rubric the Capstone uses):

• [ ] Task placed on the reliability spectrum with rationale
• [ ] Data sensitivity classified BEFORE submission, with mitigation named for any Confidential/Restricted items
• [ ] Tool category selected with justification (and verified against data tier)
• [ ] Four-element prompt with all elements labeled
• [ ] Complete, unedited AI output saved
• [ ] 3+ failures diagnosed with quoted text and named category
• [ ] 2+ professional standards named for this task
• [ ] 3+ revisions connected to named standards
• [ ] Final revised version produced alongside the original

What This Tells You

If you completed all 9 steps without getting stuck, you're ready for the capstone. If you got stuck at any step, go back to the relevant module before attempting the capstone:

• Stuck at assessment → Module 2
• Stuck at prompt → Module 3
• Stuck at diagnosis → Module 3
• Stuck at evaluation → Lesson 4.4
• Stuck at data sensitivity → Module 4

Your domain expertise is what makes this workflow work. The AI generates drafts; you know what "good" looks like in your field. That combination — AI speed plus human judgment — is what the capstone proves you can do.

Bridge

You've completed the full cycle once. The capstone is the same process on a task that matters — something you actually do every week. One last step: choosing that task and setting your baseline.