How I use (and don’t use) AI to design better in 2026

Key wellness / self-care / mindset themes

• Prevent burnout & FOMO: When AI frustration shows up, stay curious and human. Be more connected, kind, funny, and unapologetically yourself.
• Don’t over-optimize perfection: For early testing, aim for about 7–8/10 quality so you can move fast without exhausting yourself.
• Protect your body during documentation: Use AI to write meeting notes/annotations in point form to reduce repetitive typing strain (carpal tunnel risk mentioned).
• Use patience in relationships: For difficult clients, practice patience and empathy—people remember how you made them feel (not just speed/tools).

Productivity & career-resilience strategies (UX design + AI)

• Bulletproof your career (AI literacy > AI avoidance):
  • Be the kind of designer who uses AI well; don’t rely on skipping AI entirely.
  • Use AI to help you design better, while avoiding workflows that degrade quality.

Double Diamond: how/where to use AI

1) Problem space (“expand the first diamond”)

• Don’t shrink discovery:
  • Reality check: teams often reduce the first diamond, causing months wasted on the wrong problem.
• Use AI to expand problem understanding:
  • For personal projects, avoid “fake” topics (like cloning an app idea that already exists).
  • Validate business viability early:
    • Tell ChatGPT you’re starting a real SaaS business.
    • Ask for market analysis, revenue model design, and revenue predictions.
    • Iterate across multiple ideas until one is strong.
• For client work: regain influence when problem scope is preset
  • Situation: stakeholders define the problem/solution; designers are told to move fast (“visualize what they said”).
  • AI response: draft outreach/emails to pull designers into discovery earlier:
    • Email #1: to the project owner to get into the discovery meeting before the PRD is finalized.
    • Email #2: kickoff meeting invite to aligned stakeholders before the problem is defined (PM, content designer, data analysts, engineers, legal, UX researchers).
    • Email #3: to UX research + PM proposing a short primary user interview plan to secure approval before design starts.
• Don’t replace primary research with AI
  • Use AI to support—not substitute—first-hand learning:
    • draft recruiting emails
    • strengthen interview questions
    • speed up preparation
• Use AI to make the problem space “real”
  • Don’t stop at a design brief (user-only) if it lacks business goals + technical constraints.
  • Use AI to draft a PRD including:
    • business objectives
    • success metrics
    • scope
    • constraints
    • dependencies
    • risks

2) Solution space (“diverge → generate → refine/validate”)

• Avoid placeholder content as default
  • Don’t rely on Lorem Ipsum; realistic content improves communication.
• Use AI inside Figma to reduce friction
  • Replace Content (Figma AI):
    • select text layers → describe what the text should say → replace
    • can replace multiple text elements at once (better than copy/paste from external tools)
  • Translate To (Figma AI):
    • translate entire frames/multiple frames to test international layouts
    • still recommend human copy refinement before launch
• Document design iterations so you don’t redo work
  • Common failure: only final versions exist; explorations are deleted/hidden/unannotated.
  • AI-assisted workflow:
    • use an AI note tool (example: Granola) to summarize meeting notes
    • in Figma: keep latest at top, one major iteration per page
    • annotate with what changed, why, and feedback source
    • often generate bullet points by prompting ChatGPT
• Prototype complex interactions faster
  • Use Figma Make to turn hard-to-prototype elements (e.g., tables) into functional prototypes.
    • Example request: “Make each table column both sortable and draggable.”
  • Works even when you don’t have a design yet:
    • screenshot the page → start Figma Make → paste screenshot → describe interaction → get an interactive prototype
  • Safeguard:
    • if you adopt an interaction from Figma Make, bring it back into Figma and clearly annotate; treat Make output as a demo until the design system/spec is followed.

New high-speed AI workflow (especially for startups / building your own product)

• Shift workflow from “low-fidelity → high-fidelity” to “prompt → prototype → evaluate → iterate.”
• Use Figma Make heavily when speed matters.
• Intentionally diverge with multiple generations
  • Same PRD/prompt, run multiple times:
    • generate 3–5 versions
    • optionally switch AI models per version
  • Goal: avoid emotional attachment—design better by having options.
• Iterate after selection
  • Pick one direction, then prompt again for improvements:
    • start with structure, then content, then visuals
• Test using a “Flight of the Bumblebee” rhythm
  • Define happy path (core 80% experience)
  • Iterate content (text matters)
  • Polish visuals to ~7–8/10 (test fast, don’t over-perfect)
  • Test fast:
    • internal tests with teammates (pretend-user)
    • best: real user testing
    • for startups: validate demand / build early waitlist

Where AI is explicitly recommended vs avoided

• Recommended
  • Expand discovery: PRD drafting, market/revenue analysis, interview prep, stakeholder alignment emails
  • Solution iteration acceleration: Figma AI Replace/Translate, Figma Make prototyping
  • Meeting + documentation acceleration: AI summaries + point-form annotations
• Avoid
  • Using AI to replace primary research
  • Skipping the first diamond (problem discovery) and jumping straight to solutions
  • Using AI output as a “final product” without bringing it back into Figma and aligning with design systems/specs

Presenters / sources

• Presenter: Aliena Cai (also mentioned as “Aliena” throughout)
• Tools/platforms referenced (non-human sources):
  • ChatGPT
  • Figma (including Figma AI features and Figma Make)
  • Granola (AI note-taking)
  • Google Translate (historical workaround mentioned)
  • Uber (example used in discussion)
  • eBay (example context)
• Program/brand referenced: Fast Track UX (Figma Partner Design Bootcamp)