Concrete workplace scenario
The manager asks what each workflow candidate needs as input and what output it must produce.
The trap is defining input/output loosely. D63 must name required inputs, optional inputs, output artifact, evidence, and readiness signal.
Worked answer
Bad example first: the plausible shortcut fails
Weak example
“Input: customer information. Output: renewal-risk summary.”
What goes wrong
The weak version contains customer information, summary, no required evidence, no readiness signal.
Work-ready version: use the day rule
Required input
Blocks the workflow if missing.
Output artifact
The work product created.
Readiness signal
How humans know it is ready or blocked.
Quick decision check
Workflow I/O needs required inputs + output artifact + readiness signal
Use this rule before filling the build table. The goal is a decision-ready artifact, not a broad summary.
Required input
Blocks the workflow if missing.
Output artifact
The work product created.
Readiness signal
How humans know it is ready or blocked.
Workflow/SOP foundation week workflow trace
Before / after: what changed in your work?
Before learning
After learning
What AI-in-workflow means today
Workplace case
The worker selected a repeated renewal-risk update, but the process will drift unless the input packet and output standard are explicit.
Main learning job
Define the workflow input packet, output format, acceptance standard, missing-input rule, and review owner.
Failure mode
Breaking down steps before knowing what input is required and what output must be accepted.
Row/time guidance
Core: 3 input-output rows; Deep: improve the same 3 rows with stronger evidence, limits, and repair notes. Do not map steps until required input/output and missing-input rule are visible.
Artifact handoff
Previous artifact: D62 repeated_task_candidates.md
Next artifact: D64 step_breakdown.md
Learning checkpoint: Keep the handoff useful to the next lesson by naming evidence, limitation, owner, and stop rule.
Learning checkpoint
This page keeps Day 63 connected to the surrounding course path. Use the previous artifact as working input, produce today's artifact, and hand it to the next lesson with limitations visible.
Input/output contract example
For renewal-risk review, the input packet is account list, renewal date, latest support notes, and owner. The output is a risk summary with caveat and next action. Missing renewal date triggers a stop, not a guess.
Contract repair check
If the output asks for a risk label, the input must include current account evidence. If it does not, write a missing-input rule before adding another step.
Static card version of the worked example
Use these cards before the wide table so the example does not depend on horizontal scrolling.
Worked move
Input/output contract example
Decision standard
If the output asks for a risk label, the input must include current account evidence. If it does not, write a missing-input rule before adding another step.
Day-specific completed sample row
| No. | Workflow candidate | Required input | Optional input | Output format | Acceptance standard | Missing-input rule | Review owner | D64 step need |
|---|---|---|---|---|---|---|---|---|
1 | Weekly renewal-risk update | CRM risk fields, latest account note, renewal date | Customer success comment | Manager-ready risk summary with evidence links | Every risk claim has source and requested decision | If CRM note missing, mark unknown and request owner update | Sales ops lead | Break into collect, verify, classify, draft, review, send |
Build workflow_io_definition.md
Core: 3 input-output rows; Deep: improve the same 3 rows with stronger evidence, limits, and repair notes. Do not map steps until required input/output and missing-input rule are visible. Blank required fields create export warnings.
| No. | Workflow candidate | Required input | Optional input | Output format | Acceptance standard | Missing-input rule | Review owner | D64 step need |
|---|---|---|---|---|---|---|---|---|
| 1 | ||||||||
| 2 | ||||||||
| 3 |
Summary / decision
Downstream handoff field
Repair one weak row
Use this pass to find the row that would make the artifact look ready while hiding evidence gaps, unclear owner decision, missing exception, or weak handoff.
Repair example
Weak row: Input: customer info; output: summary.
Repair move: Add required evidence, optional context, output artifact, evidence used, and ready/block signal.
Better row: Input: current CRM renewal note; optional support ticket; output: risk summary with source; ready if source current, blocked if stale.
Choose the row you would be comfortable carrying into the next day.
Choose the row most likely to fail before AI use.
Sentence starter: The weak row is weak because ___. I will repair it by adding ___ before using AI.
Export artifact
This file is today’s work receipt. It should show the exact decision or handoff that the next day can use, not just that you filled a table.
Before you export
- Does the export distinguish required input, optional context, output artifact, and readiness signal?
- Does it avoid broad input/output labels?
- Would D64 know which steps must be broken down?
What this artifact proves: This file proves the learner can define workflow inputs and outputs tightly enough to execute.
Weak export: Input: customer info; output: summary.
Good export: Input: current CRM renewal note; optional support ticket; output: risk summary with source; ready if source current, blocked if stale.
Click Generate Markdown to create workflow_io_definition.md.