Marina Makeewa
Back

AI tools in the team

Review errors −30%, research coverage 98%, knowledge-sharing ×3.

Role
Lead of Design Expertise
Period
2024 — 2025
Company
Gazprombank

TODAY 2026

Today I'd do it this way

Rolling out AI tools to a design team is no longer about licenses and training. It's about configuring skills around the team's actual workflow + measuring baseline → new metric in 6 weeks. The selection and adoption itself compresses from 3 months to 3 weeks.

Then3 months
Now3 weeks
Skill I'm writing for this

tool-rollout-playbook

In progress

Team audit + skill configuration + before/after metrics.

Capabilities
  • 12-question audit: where the team loses time, what's automatable with skills
  • Baseline measurement on 5 metrics before rollout
  • 6-week tracker with 'continue / roll back / pivot' triggers

ARCHIVE 2025

How it was done then

Summary

Full version available in Russian. Read in Russian →

Designers were getting tight deadlines with no time for research or proper testing. I mapped where exactly time leaked and worked with the team to pick a focused AI toolset — Mobbin for UX references, Pathway for fast research, AI tools for product and graphic design.

After rollout: review errors down 30%, research coverage up to 98%, knowledge-sharing initiatives tripled. The team shifted from "ship fast" to "ship well".

What I'd do differently today

Today I wouldn't buy a SaaS stack. I'd write Claude Code skills for each designer — design-review agents reading their own Figma frames via MCP, research-prompts generating interview guides for specific products, pattern-analyst scanning flows against the design system. SaaS gives an averaged tool. Skills give designers a lever they configure for themselves in one evening.