All work

ASI Central · ESP+

The B2B platform behind North America’s $27.7B promotional products industry. I designed Print-on-Demand, built on the Cosmos design system and helped the team move from mockups to prototypes in code.

Role
Senior Product Designer, 1+ year
Team
Design team led by a VP of Design; I brought in 3 designers
Platforms
Web and mobile, 14 product areas, 2 brands
11 → 3.5 daysbrief to clickable prototype
3.4 → 1.3review rounds per feature
−73%requirement changes after sign-off
19 → 7defects per feature set

ASI has run the promotional products business since 1962. ESP+ grew through acquired modules, three generations of front-end framework and a design language that drifted every time a team shipped in a hurry. The job was to keep shipping new features without adding to the drift.

  • Distributors bought blank products in ESP+, then left to decorate them in one-click print-on-demand services.
  • One product spread across 14 areas: search, orders, quotes and invoices, CRM, collections, presentations, websites, stores, analytics and more.
  • Two brands, ESP+ and ASI Central, on one inconsistent component library.
  • Details surfaced late: missing states in review round three, limits found in QA, components that didn’t exist on handoff day.

  • Designed Print-on-Demand end to end, contributed to the Cosmos design system and helped move the team from static mockups to AI-built prototypes in code.
  • Grew the design team: ran interviews and onboarding for 3 designers I brought in, and helped backfill roles when designers and our UX researcher moved on, so delivery didn’t stall.
  • Worked in a design team led by a VP of Design, alongside PMs, front-end and back-end engineers and QA across 14 product areas.
  • Options, not one answer. For most features I brought several options to the weekly design review with the whole product team, explained how I got to each and which problem it solved, and we chose together.

POD is a fast-growing market (about $13B in 2025, ~26% a year), and we were losing the decoration step to single-purpose tools. Distributor interviews kept coming back to the same line: they bought blanks in ESP+ and printed somewhere else.

  1. Pick a POD-ready productStart from the same catalogue and order.
  2. Place artworkArtwork, text or uploads on the front and back.
  3. Choose method and coloursDecoration method and colour management in one panel.
  4. Set price and marginA price grid and margin, with no second tool and no re-entering data.

When we added fundraising to pop-up shops, stores built for one event such as a team launch, the obvious move was to show it in every existing flow. Interviews and research showed it mattered only when two conditions met at once, and that combination was rare.

Hypothesis: adding it everywhere would clutter the flow for most users, who never needed it. We kept fundraising where it was needed and left the main flows clean. Our users told us where not to add.

Cosmos 1.2 powers both ESP+ and ASI Central: 290+ component sets, 5,300+ variants and states and 500+ design tokens. I helped build and document its components, states and page patterns, so Search, Orders, CRM, Collections, Presentations and POD feel like one product.

Before bringing AI in, we spent six weeks cleaning up the library, so there was something worth generating from.

  1. BriefProduct, design, front-end and QA agree on states, data and a closed list of components in one short session.
  2. Claude and Figma MCPClaude reads the Cosmos library in Figma and builds only from components that already exist.
  3. Figma and GitHubOne run gives Figma frames for every state and a clickable prototype on a GitHub branch.
  4. EngineeringDevelopers keep the prototype’s front end and wire up data, auth and tests.

We kept Figma on purpose: it stays the lasting, reviewable record, and designers review, edit and sign off every generated screen. Read the full story on Medium.

  1. ResearchInterviews with distributors and suppliers.
  2. HypothesisWhat we believe and how we’ll know.
  3. CritiqueDesign critique rounds with the team.
  4. PrototypeClickable, in code, on Cosmos.
  5. ValidateTest with users and stakeholders.
  6. ShipHand off to developers, then measure.

One product team, six features in a quarter:

  • Brief to clickable prototype: 11 → 3.5 days.
  • Review rounds per feature: 3.4 → 1.3.
  • Rework after handoff: 4.2 → 1.1 developer-days.
  • Defects per feature set: 19 → 7.
  • Requirement changes after sign-off: −73% (5.1 → 1.4).
  • Zero off-system components in generated UI.

How we measured: over three months across six features on one product team. The client tracked these numbers and their design leadership shared them with us; we had no direct access to the analytics. Results reflect the work of the whole product team.

POD gave distributors a reason to keep decoration inside ESP+. Cosmos gave two brands one consistent language. And the team’s role shifted from drawing screens to engineering prototypes, with Figma kept alongside for review and as a copy of every screen.

  • Cosmos design system
  • Print-on-Demand
  • AI design engineering
  • B2B SaaS

Align with design leadership early, not case by case. I worked inside a product team, while design decisions also went through central design leadership. Sometimes my team agreed on a solution and it then stalled in design review, and I worked around it one feature at a time.

Today I would set up joint sessions from the start with the design lead and my product manager, agree on what good looks like and how decisions get made, and bring hypotheses and options to my own team earlier. Then the team spends its energy on users, not on alignment.

Next caseUrideTwo native apps designed from zero: 2× completed driver sign-ups and ride booking from 2:40 to 0:55.

Need someone to lead
your
design team?

I’m looking for a Head of Design role, remote from Spain. English, Ukrainian and Russian, intermediate Spanish.