Framework · AI & design · 2 min read
AI adoption ladder
A four-rung scale for where a designer is with AI on one specific task, so you can coach the next step instead of pushing “AI adoption” in general.
-
Explore
- AI does
- Small, low-risk jobs you try out
- You do
- Check everything and learn what it can and can't do
- For example
- Rename layers, summarise meeting notes
-
Assist
- AI does
- Writes a first draft
- You do
- Edit, correct and decide what stays
- For example
- Spec drafts, research guides, UI copy
-
Co-create
- AI does
- Generates options and prototypes with you, inside the design system
- You do
- Steer each round and choose between options
- For example
- Layout variants, interactive prototypes
-
Orchestrate
- AI does
- Runs several steps as agents: spec, prototype, code, review
- You do
- Set the goal and the checkpoints, review at each one
- For example
- Spec-to-prototype pipeline with review gates
At every rung, the designer owns the outcome.
Use it when
- You are planning how your team will adopt AI and need a starting point that isn't "everyone use tool X".
- A designer is stuck, sceptical or anxious about AI, and you want a concrete, low-pressure next step.
- You need to choose who can pilot a new agentic workflow, and for which task.
How to use it
- List the real tasks. In a 1:1, list the 5–8 tasks the designer does every week or sprint: research synthesis, specs, UI copy, prototyping, handoff QA. Output: a task list for this person. About 15 minutes.
- Place each task on a rung. Ask two questions per task: "What does AI do in this task today?" and "What do you still do yourself?" Place it by what happens now, not by what they could do. Output: one row of the map below, filled in.
- Choose one task to move up one rung. Pick a task that is frequent, low-risk and where the designer is curious. Never jump two rungs at once. Output: one goal for the next four weeks.
- Match your support to the rung. Each rung needs a different style from you, the same way Situational leadership works. Output: what you will do, written next to the goal.
- Review it monthly. Revisit the map in 1:1s. Share patterns with the team or the AI champions network, never individual positions. Output: an updated map and the next goal.
Worked example
A fictional mid-level designer, mapped in a 1:1. Click a rung to move a task and see what you'd do next as their lead.
Task
ExploreAssistCo-createOrchestrate
Watch out for
- Turning it into a score. The ladder is a coaching tool. If it shows up in a performance rating, people will stop telling you the truth about how they work.
- Labelling people. "She's an orchestrator" hides the tasks where she is still exploring, which is where she needs you most.
- Moving sensitive work up first. Check the data rules in AI governance and data protection before a task that touches personal or confidential data goes past Assist.
- Dropping the review at higher rungs. More autonomy for AI never means less accountability for the designer. The review points move; they don't disappear.
Template
Copy this into your 1:1 doc and fill one per designer.
Designer: Reviewed on:
Task | Rung today | Next rung | My support | Check in
----------------------|------------|-----------|------------|---------
Research synthesis | | | |
Specs & documentation | | | |
UI copy | | | |
Prototyping | | | |
Handoff QA | | | |
Goal for the next 4 weeks:
- Co-IntelligenceEthan MollickBook · 2024 (Amazon)
- How AI Literacy Shapes GenAI UseNN/g (external site)
"See on Amazon" links are affiliate links. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.
Saved only on this device.