CASE STUDY

Eight teams, four products, one way AI behaves

I’ve spent years designing foundations that other teams use to make experiences consistent. RingCentral had many different products. They ran deep, each built by a different, siloed vertical team.

But users don’t work in silos. Their day runs across phone, video, messaging and contact center. That horizontal gap is where I thrived.

So when AI exploded across every product at once, I built the interaction framework underneath AVA, RingCentral’s flagship AI assistant. It let eight design teams across four products build AI at scale without losing quality.

  • 8 teamstrained over 20 designers
  • 4 productsincorporated the framework
  • ShippedAVA launched in November 2025, desktop and mobile
  • Built to scaleAI-native framework with rules, 30+ specs, a 33-page wiki, and skills files that teams’ agents built from.

Highlights:

  • Core behavior rules and the co-pilot page pattern.
  • AI branding framework v1, in all four products.
  • The utility rail in RingCX, adopted by Phone, SMS and Video.
  • Agentic onboarding, aimed at one of our biggest call drivers.

First. What matters.

Background

Systems thinking was how I'd always worked at RingCentral. One of the first things I did at RingCentral was realize we had no set of shared rules for how the product communicates feedback back to the user, so I wrote the framework patterns, matching severity to friction from a quiet toast to a blocking dialog. I found myself designing a system that a lot of chaos could flow through and still come out consistent. I did the same thing with a first-use framework, creating standards for discoverability and reengagement that multiple teams could use.

Please enter password to view full project. If you need access, email me.

Home