Specialized solutions to your content problems

Situation

I joined as the first and only embedded content designer on a team with 12 product designers, 7 PMs, and 30+ engineers. The product had been scaling for a few years without someone keeping track of the words.

Problem

Two years of content debt had built up across every surface. Inconsistent terminology, jargon in microcopy, and generic error messages with no context on what went wrong. I had to get the lay of the land to find what was broken and make a case for fixing it.

What I did

I ran a full audit across all product surfaces, meta experiences (onboarding, spotlights, temporary messages), and 60+ support documents, prioritizing each one by severity and effort. Then I launched into a support doc uplift program to work through the highest-priority content fixes.

Result

The audit surfaced 30+ text string fixes in the product, 6 larger UX/interaction issues, and ~20 signals for new content patterns. The support doc uplift program improved helpfulness scores by 18%.

Situation

Atlassian's platform apps—Goals, Teams, and Projects—had no shared framework for object types. Every layer of each app used the same name as the app itself, which was fine until the apps needed to support different kinds of objects to serve user needs.

Problem

Enterprise customers were requesting OKR support in the Goals app, but the object model couldn't accurately support a bundled Objective and Key Result relationship. Whatever was built for Goals had to be flexible enough for Teams, which had its own specific type needs, with more apps to follow.

What I did

I built a new object type system across the platform, including a new dependent object type called a “Success measure” to mimic the relationship between Objectives and Key Results. I defined the naming patterns and hierarchy that resolved the model confusion, and established the foundation that Teams would use for its own type system.

Result

In the first month after launch, users created ~1,500 measures per day, and the enterprise clients we spoke with said they'd be able to move forward with adoption. The system laid the naming foundation for both Teams and Projects' own type systems.

Situation

The Goals and Projects apps had a “Risk” feature that let teams log risks—but that's about it. Descriptions only, no lifecycle or mitigation features. You couldn't even mark it as resolved or mitigated.

Problem

Most AI integrations at the time were adding chat or text summaries to just about any surface. I wanted to try extending an existing feature using AI and automations, rather than going through the roadmap process and building it ourselves.

What I did

I RAG-trained an agent on 15 years of real, internal Atlassian risk documents to establish the right structure and content expectations. I then designed the output template for usability, surfacing the critical information first. Then I built the full automation workflow in Studio: analyze a comment, generate a structured Confluence page, and reply in thread with a link to the page.

Result

A working proof of concept with a live demo. The agent produces structured, actionable risk summaries from unstructured comments. The automation trigger had one unresolved bug at the end, but agent design, template IA, and evaluation rubric were complete.

Situation

Utility customers were enrolled in Time-of-Use electricity plans that charged more during peak hours (usually from 4-8pm) and less the rest of the day. Utilities needed customers to shift their usage, but first their customers had to understand how their new rate plan actually worked.

Problem

Before we could motivate behavior change, we had to focus on time-of-use comprehension first. Customers needed to see their own data, understand what it meant, and build new behaviors over time. The content had to make a complex billing concept feel simple and personal.

What I did

I wrote a three-part email series covering the introduction, weekly update, and monthly recap, using progressive data disclosure and light gamification to build understanding and reinforce behavior over time. Each email led with the most actionable insight first, anchored to the customer's own usage.

Result

The program drove a 1-2% peak load shift across all five pilot deployments. At one utility, digital engagement increased by 60% and the emails saw 3x the opens and 9x the clickthroughs compared to industry benchmarks. Recipients were 17% more likely to recall the program and 16% more likely to correctly identify their peak hours.