Yingying Zhang
AI-native product direction

AWS IoT Experience Vision

2024–2025 · North Star UX Vision Co-Lead

It started as an unowned effort to make today's workflow easier. I pushed it forward: from asking customers to stitch 18+ services together by hand, to letting them describe a goal and have the system assemble it.

AWS IoT had 18+ services, customers took a year or more to reach production, and the work had no owner, no dedicated resources, a reorg pause, and AI initially out of scope.

At a glance
The problem18+ services, a year or more to production
ConstraintNo owner, reorg pause, scope kept narrow
My roleInitiated it, co-led the North Star
Strategic moveToday's workflow → AI outcome-building
OutcomeLeadership support, dedicated PM
The project
The situation

Customers were struggling to stitch 18+ services into a working solution. As one of them put it, "Do not ship your org chart to us."

One working solution
Eighteen-plus separate services, scattered. The real work was stitching them into one working solution by hand.
The move

The original direction combined UI unification with workflow automation. It improved today's experience, but still left customers responsible for knowing the services and stitching the solution together.

Design leadership challenged us to look further ahead. If this was a North Star, it needed to show a new way of working, not just a cleaner version of today's workflow.

I reframed the vision around AI orchestration: customers describe the outcome, and the system helps assemble the solution.

SERVICE BUILDER Learn each service's API Stitch the services by hand Write the glue code OUTCOME BUILDER Describe the goal System assembles the parts Configuration generated
The reframe: the customer stops being a service builder and becomes an outcome builder.
Engineering Reusable modules Intelligent layer Works backward from the goal Go-to-market Speed to value
An intelligent orchestration layer, bridging Engineering's reusable modules and go-to-market's need for speed.
The outcome

The work changed the question for leadership: not “how do we make today’s workflow easier,” but “how do we help customers describe an outcome and let the system assemble the solution?”

I designed the orchestration layer to work backward from the customer's goal, reusing Engineering's existing modules instead of asking for new ones, so go-to-market could reach customer value faster. Then I packaged the vision so my design partner could carry it forward during my leave.

The vision won leadership support and dedicated PM ownership. It survived a reorg, a long pause, and my absence because the story was clear enough for others to keep moving.

Initiated found the gap Paused reorg, AI out of scope Pivot to AI orchestration Handed off built to run without me
It survived a reorg and a long pause, and my leave.