Atlassian
Product Design Manager · 2026
I led the team responsible for the modernisation of Jira's work item. The aim is to update a twenty-year-old product by giving it a user interface consistent with the Atlassian Design System, adding new AI features, and organising the system around what customers really want to achieve.
As Product Design Manager, I led the effort to modernise Jira's work item, a surface that millions of teams use daily. The objective is to create a modern user interface that conforms to the Atlassian Design System, reorganise it around customer ‘jobs to be done’, and incorporate AI where it truly helps, all while ensuring existing workflows are not disrupted.
I led a team of six product designers to create a new design vision for the work item surface. I was ultimately responsible for the quality of design output, and worked with the Heads of Product and Engineering to ensure the cross-functional team delivered a high-quality product. To keep everyone aligned, I organised a leadership design sprint, used a common 'north star' to guide daily decisions, and established simple team rituals. I consulted stakeholders to include AI only where it benefited users and personally designed some of the most important user journeys.
The project is ongoing, but the platform's performance is now 30% faster. The work item aligns again with the Atlassian Design System and has been reorganised around jobs-to-be-done. AI meets real customer needs as we approach ten million agent invocations a month. Most importantly, 6 designers, 7 product managers, and over 80 engineers are working towards the same vision.
Jira's twenty-year history was evident throughout the user experience. The continuous addition of features has diluted its visual style and moved it away from the Atlassian Design System. Instead of organising by what customers need to do, it follows the order features were developed. Its speed and performance have fallen below current expectations.
In practice, we modernised the interface to align with the Atlassian Design System, reorganised it around customers' ‘jobs to be done’, and made it considerably faster and easier to use. The problem was that millions of daily workflows rely on this product, so I faced a difficult task. Ideally, we would change everything without disrupting anything.
Halfway through the year, the success metrics changed. Atlassian shifted its business model from charging for paid seats to ‘usage-based pricing’ based on AI. The key metric moved from monthly active users to AI agent invocations, targeting ten million per month. This change broadly impacted the product, affecting the user interface architecture and features. The work item became a place where people and AI agents work together, influencing all decisions about where AI should be included.
Each day, I was responsible for getting the best performance from six designers at various levels. I held regular one-on-ones, provided frequent feedback, and assigned tasks matching their strengths so they could produce their best work and continue to grow. Where possible, I let them lead their own projects beside their product and engineering counterparts and make decisions without routing everything through me. Believing in 'servant leadership', I ensured each team member was happy, healthy, and had what they needed to succeed. I gave advice, offered continuous feedback, and only intervened when necessary, avoiding micromanagement.
I was responsible for the design strategy of the work item as well as the quality of design’s output. I helped define specific design principles the team could work towards and maintained the standard through design critiques and reviews. With many parallel projects, keeping the standard consistent was key so all releases felt like a single, holistic product rather than many separate ones bolted together.
Making progress on a project of this scale involved keeping stakeholders aligned. I worked closely with the Heads of Product and senior PMs to develop the product strategy, ensuring the design direction supported the product's goals. I monitored the team's capacity, raised trade-offs early, and kept design decisions moving to meet delivery deadlines without lowering quality.
I went beyond my own stream by participating in regular design reviews with design leadership, sharing our work for feedback, and learning from other teams' insights. I also helped ensure our practices stayed aligned with standards and patterns used across the wider design function.
I ran a design sprint with the Heads of Product and Engineering before designing features to agree on what we would launch, then turned this into a shared roadmap. We converted the 'iconic 15/10' ask from executive leadership into specific quality standards and experience principles for the team. We based the restructure on customer conversations and usage data to align the new information architecture with real jobs-to-be-done rather than stakeholder opinions.
Together with lead designers, product, and engineering teams, we developed a single, ‘north star’ vision so everyone could understand and take ownership of it rather than just following it. The vision became a daily decision-making tool, not just another artefact to be forgotten. It resolved trade-offs that might have led to lengthy debates and ensured six teams and eighty engineers worked towards the same goal.
A few established rituals such as critiques, design reviews, and decision logs made it possible for everyone to see and understand the design rationale. We divided the surface into separate projects matching each designer's strengths. We used shared patterns and the north star to ensure these parallel efforts combined into a single coherent experience.
In this new AI age, technology and customer expectations are advancing rapidly. This sometimes caused design work to be wasted as feature requirements and industry standards kept changing. To address this, I helped the team accept that this was normal and that nothing is truly wasted since all we build improves on what came before.
Given the aim of ten million invocations, the natural tendency was to incorporate AI into all possible areas. At the same time, we were learning from one of our competitors. Microsoft had introduced Copilot to almost every surface, such as Notepad, the Snipping Tool and the floating buttons on top of the canvas, and although engagement increased, so too did the criticism as people felt that the software was becoming less helpful, and by March 2026 Microsoft had publicly eliminated a number of unnecessary Copilot entry points in its main applications.
Initially, AI usage was more about generating clicks than providing value, and the invocations metric could be manipulated through clutter. So, instead of spreading AI across the interface, we compared each entry point against a real customer's job, shortlisted them, and deliberately reduced it. Now AI appears only where a job requires it, not everywhere we could place it. We chose ‘simplicity over proliferation’.
A common set of principles guided how people and AI should work together. The interactions should be proactive, using our knowledge of the individual, the team and the company to provide help at the right time, and they should remain part of the natural flow so that the experience feels natural and involves as little friction as possible. They should be dynamic, allowing people to switch between text, audio, video and visual formats, and they must always be inspectable so that anyone can take a look underneath the surface to either understand a result or access the details beneath it. They also need to be collective, linking individual contributions to team knowledge while taking account of permissions and enabling smooth transitions between people and agents, in a manner that still clearly reflects Atlassian.
The project is still underway with a launch set for October 2026. It includes eight core improvements that form a unified experience. A new quick-create process makes it easier to capture work. All work item elements have a consistent visual appearance throughout Jira, and the content hierarchy is redesigned around customer needs. We introduced a simpler empty state with small pleasant surprises, used AI extensively but with restraint, and improved load times and performance by 30%.
The 'iconic' experience we were striving for was defined by the fine details. It was not one major decision but the result of a thousand small design choices that met the same high standard. Each element has its place and the information hierarchy is reconstructed based on customer needs so the eye moves through content in the correct sequence. White space is part of the structure, allowing the dense tool area to breathe. The work item uses a single icon language from the Atlassian Design System. The palette remains neutral by default and uses colour only when appropriate. Illustration is used sparingly to add warmth to empty states and first-time experiences rather than for decoration.
The project is ongoing with a clear direction. Performance has improved by 30%, and the work item is aligned with the Atlassian Design System and consistent across Jira. Content is organised by jobs-to-be-done rather than feature order. AI focuses on entry points matching real customer jobs, aiming for ten million agent invocations per month without unnecessary clutter. What I am most proud of is that six parallel projects and eighty engineers have converged on a single shared vision and produced a holistic set of high-quality user journeys.