API Center

Since 2025 I have been design lead for Azure API Center, a sub-service in the Azure API Management ecosystem focused on helping enterprises catalog not just APIs but new AI assets like MCP servers and agent skills. I led the design of two connected surfaces: the open-source API Center Developer Portal, where developers discover and consume these assets, and the Azure Portal registration experience, where teams register and configure the assets they publish. Beyond design, I authored and shipped the front-end implementation of the developer portal redesign directly.

As my project is under NDA, I can't disclose full project details. If you are interested in learning more, please reach out to me at clairedycat.kantner@gmail.com.

Duration

Jan 2025 - Ongoing

My role

Product design lead, design + front-end implementation

Problems with our existing developer portals

Only built for APIs. The Azure portal interface offers traditional APIs like REST, but lacks infrastructure to support a modern AI developer's tools such as MCP servers, skills, agents, plugins, or models.
Inconsistent, dated UI. Detail pages diverged in layout, with some content living in slide-out drawers, others in full pages. It lacked a shared card system, dark mode, and responsiveness.
Hard to actually consume. Endpoints weren't easily copyable and install actions were buried, so "I found it" rarely led cleanly to "I'm connected."
Registration gaps. Publishing new asset types through the Azure Portal surfaced missing or confusing fields, making it hard to register an asset correctly the first time.
‍

How might we make it effortless for an AI-native developer to discover, trust, and consume any kind of asset through one consistent, modern portal?
‍

Design principles
‍

Discoverable
One portal to find every asset type. A unified discovery homepage and consistent iconography aligned with Microsoft's Fluent design system.

Consumable
Optimize for the moment after discovery. Copyable endpoints, clear install actions, and remote/local guidance turn "found it" into "connected" in seconds.

Extensible
Design a system that scales to new AI-era asset types. Shared patterns for tabs, badges, documentation, and quality assessments so the next asset type slots in instead of starting over.
‍

ai assessments

AI often makes mistakes and not all assets in a catalog are trustworthy. I designed and implemented an AI quality score view to help developers decide which agent skills in the catalog are vetted.
‍

prototyping with ai

My biggest learning experience was embracing AI coding tools to empower my design workflow. For example, I used Copilot to code my own prototyping settings to quickly compare UI options in context. Once I was happy with a direction, I could easily push designs directly to the codebase, saving tons of back-and-forth we had before during engineering handoff.
‍

Project impact

Shipped designs to our 2,600+ enterprise customers, including Fortune 500 companies.
+117% Standard SKU adoption growth year over year.
6 new asset types offered beyond traditional APIs.
20+ Pull requests merged to the open-source APIC developer portal repo on GitHub

Next project