Unify was a greenfield platform inheriting the technical and design debt of products built independently. The navigation needed to hold every CloudBees product in a single, coherent model. UI patterns had diverged across product lines. And the design team — now seven people — needed a shared foundation to work from, not seven parallel interpretations of the same platform.
Solution
Led the platform-wide IA effort through four iterations of navigation architecture, converging on a modular v4 model. Standardised the high-frequency UI patterns — settings surfaces, summary tabs, entity CRUD navigation. Embedded designers into the component dashboard and key feature areas as strategic partners.
Inheriting a team, defining the direction
Head of Design — seven designers, one platform
The mandate was clear: make the platform feel like one thing. Unify was a greenfield product built against a deadline, inheriting the legacy of products that had grown independently. The first task was structural — understanding what each designer was working on, where the critical gaps were, and how to embed the team into the strategic areas the platform needed most. Designer placement wasn’t ad hoc; it was a deliberate model, aligning people to the surfaces that mattered most to the platform’s success.
001002
Building the platform spine across four iterations
Navigation
Navigation was the first and most persistent structural problem on Unify. Starting from the CDRO navigation work and iterating through four major versions, the goal was a model that could carry every CloudBees product — CI, CD, feature management, analytics — in a single, coherent shell. Each iteration answered a new question: how do org hierarchies nest inside the sidebar? How do settings surfaces generalise across products? How do summary and detail tabs standardise across different entity types? Universal search added a cross-platform retrieval layer on top. The result was a navigation system built from composable parts, where each pattern — groupings, orgs, settings, search — could evolve independently without breaking the whole.
003004005006007008009010011012013
From structured flows to AI-agentic guidance
Onboarding
Onboarding on a multi-product SaaS platform is structurally different from onboarding on a single tool — users arrive with different goals, different levels of DevOps familiarity, and different starting points. The v2 onboarding sketches established a structured, task-oriented flow that could guide new users from account creation to first meaningful action. The AI-agentic onboarder pushed this further: an LLM-guided flow that could adapt to user context in real time, surfacing the right setup steps for a developer, a release engineer, or an admin without requiring them to navigate a fixed decision tree.
014015
Component visibility across the platform
Dashboards
The component dashboard was a strategic feature: a unified view across all pipeline components in a project, giving engineers a single place to understand what was running, what was failing, and what needed attention. The work followed the full arc — a live discovery session to map the problem space, mockups to test the information hierarchy, then a working prototype with loading states and interaction detail ready for engineering handoff.
