Mike Hakobyan
AI Product Design

AI product design, built on your system.

I help product teams make their design systems readable by AI — so tools like Claude and Figma MCP produce on-brand, on-system work instead of generic output.

Method

From audit to AI-assisted workflow

The same process I ran for EatApp, where 150+ components were tokenized and connected to AI.

01

Audit.

Every screen, component and style is inventoried. Duplicates, one-offs and drift are mapped before anything changes.

02

Tokenize.

Colour, type, spacing and components become named tokens — the vocabulary both humans and AI tools share.

03

Document.

A DESIGN.md describes the system's rules in plain language, so AI assistants generate on-system work instead of guessing.

04

Connect.

Figma MCP and Claude are wired into the workflow, so explorations start from real components, not from scratch.

05

Review.

AI output is explored widely, then reviewed by a human before anything ships to design or engineering handoff.

Tools
  • Figma MCP
  • Claude
  • Design tokens
  • DESIGN.md
  • Human review
What you get

A system AI can actually design with

Tokenized components. Your library rebuilt as named tokens AI tools can read.

A DESIGN.md. Plain-language rules that keep AI output on-system.

A working workflow. Figma MCP + Claude connected, with human review before handoff.

Work together

Want AI designing on your system?

Bring your product to a 30-minute call and we'll work out the right shape of engagement.