Marcus Andrey Study
Ler em português Library

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.

  1. 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
  2. Assist
    AI does
    Writes a first draft
    You do
    Edit, correct and decide what stays
    For example
    Spec drafts, research guides, UI copy
  3. 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
  4. 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.

The four rungs of the AI adoption ladder. Each step up gives AI more of the work; accountability never moves.

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

  1. 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.
  2. 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.
  3. 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.
  4. 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.
  5. 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:

"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.

↑ ↓ to move · Enter to open · Esc to close