A powerful platform that spoke in many voices.
Automation Cloud was deep and capable, but its language had never been designed. Terminology drifted between teams, navigation had accreted rather than been planned, and new features arrived with whatever words their squad happened to choose.
There was no content design function to push back. My first job was to make the problem visible — and to build a practice that could fix it for good.
I catalogued the whole platform before changing a word.
I ran a comprehensive audit of Automation Cloud — every navigation item, header, empty state, and supporting doc — tagging terminology, voice, structure, and interaction-pattern gaps as I went.
The audit turned a vague sense of "things feel inconsistent" into a concrete, prioritized map of opportunity. It became the reference everyone pointed to when deciding what to tackle first, and it gave content design its first seat at the planning table.
20 product areas. 24 different global headers.
One pattern made the case on its own: the bar at the very top of every screen — the most-seen component in the product — had been rebuilt almost as many times as there were teams. Different layouts, different controls, no shared model. It crystallized that the platform needed more than one shared language: it needed a new information architecture, a unified global header, consistent page structure, and a rethought navigation.
One idea, wearing seven different names.
The deepest problem the audit surfaced wasn't a layout — it was the words. Across the product, overlapping terms competed for the same meaning, and every new feature deepened the confusion. Users couldn't predict what a button would do, and support couldn't speak the same language as the UI.
To get anything working with it, you need to be really into UiPath terminology — and still it just doesn't make sense.— UiPath customer
From a list of problems to a shared mental model.
To move the team from symptoms to causes, I started by mapping the whole experience end to end — building the platform's first customer journey map. It charted how users grow from their very first automation to running the entire show, and exactly where the language broke down at each stage.
With the journey understood, I worked across teams to model the system itself — defining the core objects users actually work with, their attributes, and the actions taken on them. If we could agree on what a thing was, we could agree on what to call it.
The nouns of the product.
From there I defined the core objects users actually work with — along with their attributes and the actions taken on them. This object model became the backbone for naming and contextual navigation: a single source of truth the whole team could point to when deciding what something was.
The audit kicked off a wave of work — starting with the homepage.
With shared evidence and a shared model, the team could finally act. The most visible result was a top-to-bottom redesign of the Automation Cloud home and its navigation — the first thing every user sees.
The old home buried the product behind a long, ungrouped sidebar and a promo-heavy dashboard. The redesign cut the clutter, grouped the navigation around the object model, and put an AI-first "what can I help you build?" moment front and center. A clearer global header and a refreshed iconography set followed — so visual and verbal language finally agreed.
Solving the audit's biggest finding: terminology.
One governed vocabulary.
The worst finding deserved the platform's biggest fix. I led the creation of a shared terminology system (internally, "One Platform Language"): a single canonical name for every core object, plain-language definitions, decision rules for when to use each term, and governance so the language would hold as the platform kept growing.
Then I pressure-tested it where users actually meet language — the menus, the navigation, the moment of creating something new — iterating the naming until each choice was unambiguous.
A canonical name for every core object, with definitions and usage rules teams could actually look up.
Navigation, global header, menus, and documentation began drawing from the same dictionary instead of inventing terms.
Governance and a shared object model gave every future feature a place to find the right word — not coin a new one.
Turning the standards into a tool every team could use.
A terminology system only helps if people actually follow it. So I co-created the UiPath Content Assistant — an chatGPT-based AI writing partner trained on UiPath's voice, terminology, and style guide.
Instead of a content designer reviewing every string, the standards were encoded into a tool that drafts on-brand tooltips, errors, and notifications on demand — keeping language consistent at scale and freeing the team for higher-impact work.
Same prompts, different training. The Assistant's output is shorter, sentence-cased, and on-voice — and it ends with the clear next step a generic model leaves out.