01 / Case Study — 2025
Agent Hub
I led Learn and Adopt: helping enterprise admins understand AI and plan its activation within Copilot Service Admin Center.
Release milestone: Agent Hub entered public preview in August 2025.
Selected prototype screens, not the complete experience. UI values are part of the mockups, not measured project results.
The brief
Enterprise admins wanted to adopt Copilot and AI agents, but documentation, governance guidance, and activation steps were disconnected. The brief was to bring understanding and action into one admin experience: learn what AI does, understand its requirements, and plan a rollout.
Learn: context before action
My Learn work brought technical and governance guidance into the hub rather than leaving admins to assemble answers across separate sources. The landing page connects AI learning resources with entry points for rollout planning, credit calculation, and agent insights.
Adopt: make rollout deliberate
I designed Rollout Manager as a staged activation flow, separating the choice of agents and capabilities from their configuration. The stepper keeps progress and the later review-and-publish stage visible.
1. Choose what belongs in the rollout
Agent selection and capability checkboxes make rollout scope explicit before detailed setup. Admins can choose which capabilities to include rather than treating an agent as an all-or-nothing switch.
2. Check readiness and configure behavior
For case follow-up and closure, the flow surfaces prerequisite status and rule selection together. It keeps rule selection in context and links to rule editing rather than embedding another full editor. I worked with engineering to align the experience with existing systems.
Choosing the scope
I reviewed legacy work, reframed the problem with PMs, and compared how other platforms centralized guidance and controls. Many answers to admin concerns already existed; the opportunity was to connect them to the adoption journey.
We started with Learn, Adopt, Track, and Verify. We deferred Verify from this phase and renamed Track to Measure, giving the team a shared three-pillar structure.
I led design
Learn
Make technical and governance guidance accessible within the hub.
I led design
Adopt
Guide phased activation through agent selection and configuration.
Supporting contribution
Measure
I supported alignment with the wider hub, not primary design ownership.
What the work enabled
Agent Hub reached a public release milestone, alongside internal prioritization and recognition of the Adopt experience.
Public preview
Agent Hub entered public preview in August 2025.
Leadership
The offsite demo led to prioritization and a green light for release, with Adopt called out as a standout experience.
Feedback on the approach
Leadership team feedback
Leadership highlighted change management as a major customer challenge and valued giving customers a way to try AI agents.
“The product thinking here is just exquisite.”
Public preview feedback
“The Rollout Manager is surprisingly smart. It lets you phase deployments by team, geography, or use case — so you can scale with some thought, not just speed.”
These are qualitative reactions to the work and preview, not measured adoption or business results.
06 / Reflection
This project reinforced the value of involving engineering early and making scope decisions explicit. My focus was not just organizing information, but connecting what admins need to understand with the decisions they need to make.