June 2026

Samantha Speights

End-to-End AI Design to Development Experiment

Methodology

  • Structured 6-week progressive pilot with EUX designers paired with an engineering partner

  • Created a dedicated experimental repository to give designers a safe, controlled environment to work directly from production code without impacting live products

  • Weekly qualitative feedback collection sessions led by the Lead Researcher

  • Participants evaluated two distinct design approaches: Figma-to-VS Code sync (sending designs back and forth between tools) and designing directly in VS Code without Figma — testing both paths to understand where AI adds the most value

  • Findings synthesized across 6 theme areas each week: setup & onboarding, AI design in VS Code, Figma ↔ VS Code sync, testing/commits/PRs, prior AI experience, and engineering partner sentiment

Hypothesis

EUX designers who work directly from production code using AI-assisted tools will experience measurably fewer handoff errors and faster iteration cycles than those using traditional Figma-to-developer handoff workflows. If designers can fine-tune front-end code directly with AI support, the UI design phase will achieve a 30% efficiency gain by end of FY26.

Pilot Structure

Phase 1 — Getting Started (Weeks 1–2): Orient participants to pilot goals, tools, and workflow expectations; install dependencies, configure local environments, and gain repository access

Phase 2 — Design & Prototype (Week 3): Complete AI-assisted design tasks using generate_figma_design and use_figma; practice sending changes between VS Code and Figma

Phase 3 — Create & Share Prototype (Week 4): Generate and run tests, commit changes, push branch, and make prototype shareable with the team

Phase 4 — Quality Analysis & PR (Weeks 4–5): Rerun tests after changes, commit and push, open a pull request

Phase 5 — Pull Request Review (Weeks 5–6): Pair with engineering partner, address feedback, and update branch as needed

Phase 6 — Ship It (Week 6): Merge pull request to main and trigger deployment to production

My Role

Led research efforts, facilitated weekly feedback collection sessions, synthesized findings throughout the pilot, and supported the overall execution of the project

Current State Research

Key Findings

The Research

Before the pilot launched, 12 EUX designers were interviewed about their current design-to-development workflow. The biggest issues that came up were: not being able to find up-to-date design files when joining a project, too much back-and-forth with developers during handoff, and not enough time to produce quality work. These pain points directly shaped the focus of the pilot.

Methodology

  • Structured 6-week progressive pilot with EUX designers paired with an engineering partner

  • Created a dedicated experimental repository to give designers a safe, controlled environment to work directly from production code without impacting live products

  • Weekly qualitative feedback collection sessions led by the Lead Researcher

  • Findings synthesized across 6 theme areas each week: setup & onboarding, AI design in VS Code, Figma ↔ VS Code sync, testing/commits/PRs, prior AI experience, and engineering partner sentiment

The Experiment

Results

What’s Next?

  1. Provide onboarding documentation to reduce initial setup friction for EUX Designers

  2. Expand pilot to additional designers and product teams across EUX

  3. Test making live UI changes to the product during in-store usability testing to evaluate real-time design iteration in a production environment