Frame the design decision
Redesign the work system, not simply a task. Identify the human value that must be protected and the handoff that creates the next decision.
Pathway 2 · Lesson 0375 min
AI Workflow Redesign
Mapping and reimagining human-AI work systems
Learning Objective
By the end of this lesson, you will be able to: Map a current workflow, identify AI augmentation opportunities, and design a future-state workflow with appropriate human-AI task allocation.
40%
of current workforce skills will need to change by 2030
Engage
"The question isn't 'will AI change how we work?' It already has. The question is: are you designing that change intentionally, or letting it happen by accident?"
Explore
How organisations are intentionally redesigning workflows around AI capabilities — practical examples and frameworks.
Explain
The Four-Layer Task Architecture
Every task in your organisation can be classified into one of four layers:
Layer 1 — Fully Automated: AI does it entirely. No human involvement. (e.g., data formatting, scheduling, routine notifications)
Layer 2 — AI-Augmented: AI does the heavy lifting; human reviews and approves. (e.g., first-draft writing, research synthesis, initial analysis)
Layer 3 — Human-Led, AI-Assisted: Human does the work; AI provides support on demand. (e.g., complex negotiations, strategic decisions with AI research support)
Layer 4 — Purely Human: No AI involvement. (e.g., sensitive conversations, ethical judgments, relationship management)
✅
Design the Handoff Points
The most important design decisions in AI workflow redesign are the handoff points — where AI hands back to humans. These moments need explicit design: What triggers the handoff? What information does the human receive? What decision do they need to make? Poorly designed handoffs are where AI workflows fail.
Elaborate · Worked Example
Workflow Redesign: Content Production
Context
A marketing team produces 20 blog posts per month. Current process: brief → research → draft → edit → review → publish. Takes 8 hours per post.
The Situation
Current state mapping reveals: Research (2h) and first draft (3h) are high-volume, repeatable tasks. Editing (1.5h) requires human judgment. Review (1h) requires brand knowledge. Publishing (0.5h) is mechanical. Future state design: - Research: AI synthesises sources in 15 min (Layer 2) - First draft: AI generates from brief in 20 min (Layer 2) - Edit: Human edits AI draft — 45 min instead of 90 (Layer 3) - Review: Human with AI checking brand consistency (Layer 3) - Publishing: Automated (Layer 1)
The Outcome
Time per post: 8 hours → 2.5 hours. Quality maintained because human judgment is preserved at the right stages. The handoff points are explicitly designed.
Try It · Reflect & Apply
Map a workflow: Choose a process in your work that involves 4-6 steps. For each step, classify it into one of the four layers and describe what the human-AI collaboration looks like. Where are the handoff points?
Your notes are saved in your browser for this session.
Evaluate · Knowledge Check
1 / 3In the Four-Layer Task Architecture, what characterises a 'Layer 2 — AI-Augmented' task?
Design Practice Studio
Workflow Redesign Studio
Create the artefact
Build a before-and-after service blueprint: task step, user action, AI action, human decision, handoff evidence, exception route, and success metric.
Test with evidence
Run the future-state workflow with one realistic case. Observe delay, confusion, rework, and whether the human understands what they are accountable for.
Spaced revisit
At day 14, compare actual cycle time and quality with the baseline. Keep the human touchpoint only where it protects judgment, trust, or outcome quality.
External video references
Keep Learning on YouTube
Open in a new tab
Ready to move on?
Mark this lesson complete and continue to "Designing for Trust & Explainability".