Leena AI
Leena AI's Enterprise Design System
Built the design backbone for Leena AI's virtual assistant product, tokens, components, and enterprise data patterns unified into one system so design delivery time dropped from 10 days to 4.
Role
Lead Product Designer
Timeline
3 Months
team
3 Designers, 1 Engineer
Scope
Design System, Templates, Documentation
Final Results
Design delivery time dropped from 10 days to 4
Dev handoff got 2X faster
Why this mattered
By late 2023, Leena AI's design team had grown fast, and the product was starting to show it. Every new hire was rebuilding basic components slightly differently.
Dev handoff had turned into a guessing game over tokens and formats with every new screen.
I put together a small team to fix this properly, and we set out to build a design system that could actually hold up as an enterprise-grade virtual assistant product.
The problem
No shared library existed, so consistency depended entirely on individual memory as the team scaled
Dev handoff was slow, engineers had to reverse-engineer intent from every mock
Some of the product's most basic components didn't exist yet for enterprise use cases, so every new screen held itself together loosely
Building the system
1. Foundations
Color styles, spacing scale, elevation, and a full typography system, all tokenized, so new components inherited consistency automatically instead of being rebuilt by eye each time.
Base UI Kit
Buttons, dropdowns, text fields, checkboxes, chips, toggles, tooltips, breadcrumbs. Built directly with the designers using it day to day, so it reflected shared decisions rather than one person's visual preference.
Chat & Bubble System
The part that actually makes this a virtual-assistant system rather than a generic UI kit: 15+ bubble variants and multiple expandable states, covering everything from plain text to multi-step interactions and multi-agent handling.
Application Components
Modals, notifications, file upload, date pickers, welcome headers, and the OS and CTA components enterprise deployment actually needs, the pieces a real screen requires that a demo never shows.
Enterprise Data Patterns
A reference library of table variants and filter configurations, so future dashboards could pull from existing patterns instead of designing tables from scratch every time.
Documentation
Every component documented with its states, interactions, and edge-case guidelines, so engineers could open a component and build exactly what it was meant to be, without guessing.
The system came together in one working interface, foundations, core components, chat bubbles, and enterprise tables applied to a real screen rather than shown piece by piece.
Results
Before the system existed, every new screen started from a slightly different set of assumptions, and dev handoff meant re-explaining intent every time. After it shipped:
Design delivery time dropped from 10 days to 4
Consistency stopped depending on individual memory, since shared components carried it
Dev handoff got faster, engineers finally had documented specs instead of reverse-engineering every screen
The product started to read and feel like one consistent enterprise system instead of patchworked screens
Looking back
Leading this taught me as much about delegation and review as it did about components and systems. It worked because I delegated ownership across designers and reviewed for consistency, not because I built every piece myself.
Enterprise data patterns like tables and filters are easy to skip over when sequencing a design system, but building those early meant the less glamorous part of the system was solid before anyone touched the flashier chat components.
