Summary of "7 UI/UX Design Skills I’m Learning In 2026 | Saptarshi Prakash"

Core thesis

By 2026 AI will automate much of the execution (layouts, copy, prototypes). The competitive advantage for designers will be judgment — taste, decision‑making, storytelling — plus the ability to use AI without producing generic results. Recommended workflow: use AI to quickly produce a first, editable implementation, then apply human craft (refinement, hierarchy, polish). Half prompt‑driven, half craft‑driven.

The 7 UI/UX skills Saptarshi Prakash is prioritizing

1) Developing taste

Purpose: avoid generic AI outputs; know what to keep, remove, or change.
How to practice: constantly observe strong products, reverse‑engineer their choices (spacing, typography, color, motion), and build an internal “library” of good decisions.
Product implication: faster, more discriminating edits on AI drafts.

2) Storytelling (design as a journey)

Treat interfaces as a coherent experience: vibe/tone across visuals, copy, and pacing.
Use design decisions to guide attention and unfold content; sometimes copy should dictate layout.
Benefit: clearer communication and stronger rationale for design choices.

3) Typography (core product feature)

Why it matters: typography often carries perceived product quality — font choice, sizes, weights, pairing, kerning, rhythm, and dynamic type are levers for polish.
What it teaches: hierarchy, spacing, contrast and overall visual rhythm.

4) Design fundamentals + AI‑agent thinking

Relearn UX/psych laws by feeling and testing them hands‑on (e.g., in Figma).
New consideration: design for both humans and AI agents/automations — predictable structure helps AI and often improves accessibility for people.
Practical tip: use an “agent‑ready UI checklist” when designing flows that AI will act upon.
Product focus: design true AI integrations (across e‑commerce, finance, social), not just surface chatbots.

5) The “new Figma” (design‑to‑code / design‑with‑code workflows)

Platform capabilities: Figma server/MCP integrations can connect design files to coding environments (examples: VS Code, Cloud Code).
What transfers: frames, components, variables, and layout context; roundtrip support can import generated UI back into Figma as editable layers.
Workflow impact: prompt AI to generate rough implementations tied to the product’s real components, iterate in Figma, then send PRs — accelerates first versions and keeps output aligned with the design system.
Caveat: generated code/UI is a starting point and typically needs iteration before launch.

6) Writing better prompts (prompt engineering as a design skill)

Make prompts precise: state the goal, target user, context, constraints (platform, tone, hierarchy), what to prioritize, and what to avoid.
Break problems into small parts (e.g., design a restaurant listing card instead of the whole home screen).
Repeatable structure to try:
- Goal: what to achieve
- User: who the design is for
- Context: where it appears (platform, responsive constraints)
- Constraints: brand rules, accessibility, performance
- Priorities / avoid: what to emphasize or exclude
- Deliverables: formats, number of options, success criteria
Emphasis: prompt writing is a repeatable human advantage — know what to ask for and how to refine outputs.

7) AI image & video generation (scale and consistency)

Skill shift: the value is not a single great image but defining a stable visual style that can be reproduced across many assets (50–500) with consistent color, lighting, angle, and detail level.
Tool strategy: pick models/tools by use case (concept exploration vs. production consistency, audio sync, multi‑shot generation).
Examples of tool categories mentioned: broad creative pipelines (images/video/audio/vector), text→video stacks with frame control, multi‑shot synchronized video models.
Focus: learn each tool’s strengths, build style systems, and scale assets reliably.

Actionable guidance & workflow recommendations

Use AI to produce a fast, editable first pass; then apply human judgment to refine and polish.
Practice fundamentals actively: experiment in Figma and test deliberate deviations from UX laws.
Build real products — small apps, AI tools, and workflows — to practice integrating these skills end‑to‑end.
Be honest about gaps, iterate quickly, and keep shipping; top designers will be those who think clearly and keep delivering.

Resources & features called out

Agent‑ready UI checklist (speaker references a link in the video description).
Figma → code integrations (MCP/server connections to VS Code, Cloud Code) that enable design context export and code↔design roundtrips.
Example prompts and prompt‑engineering patterns (component‑level prompts, hero copy templates) provided in the talk.

Notes about transcription accuracy

Some tool/model/platform names in the transcript may be slightly mis‑transcribed (e.g., “codec cut,” “chat GVD,” “VO stack,” “cling and van”). The intent is to compare and pick tools by use case and to adopt Figma↔code workflows — exact names may differ.