Vijay Gupta

Codifying design taste into the engineering harness.

AI design skill that turns product specs into production-ready UI using the design system.

Role
Skill workflows, rules, guidelines, and ongoing maintenance
Company
BrowserStack · Ongoing
Impact
Generated UI is 80–90% aligned with the expected output
Diagram of a rules machine: scattered UI components move through rules and guardrails and come out as a finished interface

I have been building and experimenting with agentic design workflows and systems across our product stack. As part of this, I built and now maintain an AI design skill used by the engineering harness to turn product specs into production-ready interfaces using our design system. This drastically cuts down the time from spec to working build for known cases, so deterministic screens don't have to wait on a designer and are handled with very little manual intervention.

Connect to know more about the process.

Impact & early results

  • 80–90% aligned with the expected UI

    For known cases, the generated output is already close to the intended interface, with very little manual design intervention.

  • Polished UI without direct design input

    Engineers can generate production-ready screens for established patterns using existing design system components and composition rules.

  • Faster quality prioritisation

    Visual gaps and improvement opportunities can be identified early and quickly converted into GitHub or Jira issues for prioritisation.

  • Strong early engineering feedback

    The skill is still evolving, but engineers have responded positively to the accuracy and quality of the generated interfaces.