Pilot proposal

Design AI Fluency.

A 3-week experiment for our design chapter — building hands-on fluency in AI-assisted workflows so we enter next quarter ready to embed these tools into real projects.

Equip the design team with hands-on fluency in AI-assisted workflows so we enter next quarter ready to embed these tools into real projects.

KR1

Every designer attempts daily use and completes at least 2 structured exercises (1 from Claude Design or Figma skills) — logging one finding per exercise plus a short daily note in a shared team log.

KR2

The team produces an AI Design Playbook — a living doc capturing workflows that work, ones that don’t, and ones worth revisiting — with rationale and example prompts.

KR3

Each designer ships a 3–5 min Loom of their best use case and posts it in the company AI channel — visible work, visible learning.

Week 1

Set up & first exercise

Set up accounts, install /figma-use, each pick 2 exercises including one new, complete 1 starter, and start the daily-attempts log.

Week 2

Second exercise & mid-check

Complete second exercise; capture before/after raw material for the Loom; midpoint chat to surface early playbook candidates.

Week 3

Review & ship Loom

90-min team review: walk through every experiment, agree on what goes into the playbook and why; each designer ships their Loom by end of week.

A daily-attempts log + 10+ structured findings across the team.

5 individual Looms posted in the company AI general channel — visible work, visible learning.

An AI Design Playbook grouping workflows into works / doesn’t / revisit later, with rationale and example prompts.

Evidence to anchor next quarter’s OKR — and a defensible recommendation on extending licences.

15 exercises
Brief-to-frame sprint
Figma
Write a one-sentence product brief in chat, then have the agent scaffold a Figma frame — title, sections, placeholder shapes — and discuss what was missing from your brief.
Starter
· 20 min
Design token audit
Figma
Point the agent at an existing file and ask it to list all colour and text styles, then flag inconsistencies. Compare its output to a manual audit — where did it miss nuance?
Mid
· 45 min
Variant generator
Figma
Give the agent a single button component and ask it to generate a full variant set on the canvas — sizes, states, and icon/no-icon. Evaluate how well it handles your naming conventions.
Mid
· 40 min
Competitor teardown to wireframe
AI artifact
Screenshot a competitor’s screen, ask the agent to analyse the layout decisions, then have it sketch a lo-fi wireframe on Figma canvas based on what you’d do differently.
Advanced
· 60 min
Brief as a polished doc
Doc skill
Take rough bullet-point notes from a kick-off call and ask the agent to produce a formatted Word brief. Evaluate it against your current brief template — what’s missing, what surprised you?
Starter
· 25 min
Deck rebuild exercise
Doc skill
Pull an old presentation deck, ask the agent to extract all copy, then use the pptx skill to rebuild it with restructured slide logic. Compare the two versions with your team.
Mid
· 50 min
Research PDF → insight card
Doc skill
Upload a user research PDF and ask the agent to extract key findings, then produce a one-page insight card as a formatted doc. Test whether the synthesis is accurate enough to share with stakeholders.
Mid
· 45 min
Prompt quality blind test
AI artifact
Each team member writes a different prompt for the same task (e.g. “create a colour palette for a fintech app”). Compare outputs without revealing who wrote what — debrief on what made prompts work.
Starter
· 30 min
Spec doc from Figma output
AI artifact
Ask the agent to read a Figma frame and generate a written component spec — spacing, states, accessibility notes. See how close it gets to what a developer would actually need, then iterate.
Advanced
· 60 min
Accessibility review agent
AI artifact
Share a screenshot or Figma link and ask the agent to run through a WCAG checklist — contrast ratios, tap targets, label coverage. Document where it caught real issues vs. false positives.
Mid
· 40 min
End-to-end handoff simulation
Figma
Full workflow: brief → Figma wireframe via agent → component spec doc → annotated PDF handoff. Each step uses a different skill. Debrief on where the chain breaks and what prompts helped most.
Advanced
· 90 min
Concept exploration in chat
Claude Design
Before opening Figma, run a 20-min Claude Design session for an upcoming feature. Ask for 3 distinct concept directions, evaluate them, then take the strongest into Figma. Compare the quality of the start to your usual cold open.
Starter
· 30 min
Prototype-from-frame
Claude Design
Drop a Figma screen into Claude Design and have it generate a working clickable HTML/React prototype. Test it on a colleague for 5 minutes, then note what was preserved vs. lost — interactions, hierarchy, content fidelity.
Mid
· 45 min
DS-aware UI generation
Claude Design
Feed Claude Design our design tokens and a few component snippets, then ask for a brand-new component (e.g. a notification banner or filter chip group) that conforms to the system. Score it on token fidelity, naming, accessibility.
Advanced
· 60 min
Structured design critique
Claude Design
Upload a near-final screen to Claude Design and ask for a structured critique covering hierarchy, accessibility, brand alignment and microcopy. Compare to your own critique — what did each catch that the other missed?
Mid
· 30 min