Atlassian
A company that sells workflow tools — whose own teams still lost hours to manual work.
I led UX for EZMake, the MVP of Atlassian’s internal automation platform — a concept-and-discovery engagement to reimagine how employees build automations. The problem was fragmentation: a portfolio of automation tools so sprawling that shipping one automation could mean passing through ten or more separate tools and touchpoints.
An automation company, automating by hand.
Atlassian builds the tools the world relies on to move fast — Jira, Confluence, Bitbucket. Yet inside the company, its own teams were still losing real time to the manual, repetitive work no single tool quite solved. The brief: lead the UX for EZMake, the flagship MVP of the internal Intelligent Automation Platform (IAP) — and answer the only question that matters this early: what should this thing even be?
$10B / 25k people
The FY26 ambition that made internal productivity a CXO-level priority.
~100,000 hrs
Saved by IAP in FY22 — the thesis was proven. Now: turn a project team into a platform.
“What got us here won’t get us there.” The mandate for the next chapter: stop being a group that builds automations for people, and become a platform people build on themselves.
Powerful tools almost nobody could get through.
The internal automation platform was capable — and barely used. The baseline I designed against told the story: adoption sat under one percent of the workforce, the road from first login to a first working automation averaged four months, and most people never made it there at all. The tools existed; the path through them didn’t.
These are the platform’s baseline figures — the problem I was designing against, not an outcome my work produced.
Ten-plus tools before a single automation shipped.
To cover the real range of needs, the team owned a portfolio of automation tools — each a fine answer to a specific problem. Collectively they were a maze, with none of the connective tissue to move a person through it.
Workato
Task automation & integrations.
Camunda
End-to-end business-process orchestration.
UiPath
Robotic process automation (RPA).
NLP platform
Internal build for unstructured-data & AI/ML use cases.
Opus
Observability across automations.
Mulesoft
On its way out — being deprecated.
The 122-day ramp was a story about everyone but engineers.
Engineers ramped fast and worked independently — the average hid a much steeper wall for the non-technical builders the platform was supposed to serve. Their workarounds — scripts, manual routines, one-off integrations understood only by whoever wrote them — worked in isolation but didn’t scale and didn’t travel. A new tool only wins if it beats those.
“They wanted to learn — the potential was obvious. But the training was too broad, and never actually prepared a non-technical builder to build.”— Citizen developer, user interview
Not “the organization.” Three real archetypes.
I grounded the work in around twenty existing customer interviews rather than starting from a blank page — so we were solving someone’s friction, not an average. Three archetypes fell out across one end-to-end journey — Discover, Onboard, First Touch, Build & Operate.
Consumers
Business users who want to use existing automations — but won’t build them.
Builders
The hands-on users doing the actual creating — the focus, split three ways below.
Champions
Power users who evangelize the platform and support everyone else.
Citizen Developers
Deep domain expertise, little or no code. The under-served user.
App Admins
Technically savvy, comfortable with low-code / no-code.
Engineers
Deep software background, limited business-process context.
Not another tool — a way through all of them.
EZMake framed itself as a layer on top of the tool sprawl — a single coherent journey that tied the four stages together as an abstraction layer over the existing portfolio, rather than adding an eleventh tool to it. Four moves carried someone from blank page to shipped automation:
Decision framework
Points you at the right tool for the job — so the sprawl becomes a choice, not a maze.
Starter templates
Skip the blank page and start from a working shape you can adapt.
Ship-with-confidence checklist
The checks that turn a draft into something you trust in production.
Feedback channels
Built-in loops so the team could keep learning from real usage after launch.
I turned the research into user flows and Figma prototypes — landing, onboarding, first-use, and automation-detail screens — so an abstract strategy became something a stakeholder could actually click.
One journey, four screens.
The whole EZMake experience, stitched together — an abstraction layer over the tool portfolio, not an eleventh tool. Each stage is a real comp: a user discovers the platform, gets a persona-driven onboarding, lands in a hub where they start from a template instead of a blank canvas, and ships with sample data and install steps.
Ambiguity — and keeping one user central across three squads.
The genuine difficulty wasn’t visual design — it was ambiguity. No existing product to react to, no established scope, and a user base as broad as the organization, which is really no single user at all. The work also ran across three squads with different mandates — Community (discover), Builders (onboard & build), and Governance(operate safely). Most of the job was facilitation: keeping the citizen developer central when it would have been easy to design for the engineers who were already the platform’s loudest, most-served voice. At the earliest phase, design is strategy made concrete.
At the earliest phase of a product, design is strategy made concrete.
Over roughly three months I took EZMake from an aspiration to a substantiated concept: a problem quantified against real adoption data, a set of archetypes that told us whose friction we were solving, a scope defined with three cross-functional squads, and prototypes — a redesigned hub, a standalone persona-driven onboarding, a Discover landing, and an automation-detail page — that reframed a pile of disconnected tools as a single, navigable path built around the person the old platform had left out.
This was a concept & discovery phase: the platform’s improvement goals (halving ramp time, growing to a thousand active builders) are targets, not results my work achieved — and some internal specifics are held back here pending a shareability check.