Developing a vision for AI enabled care coordination
Our Head of Product tasked me and a co-design lead with envisioning what AI-enabled care coordination at Cityblock could look like three years out. Her challenge: Be bold. If teammates recognized our current platform in the vision, we hadn't pushed far enough.
A vision built on cross-functional collaboration
We started with existing evidence, pulling insights from recent research, past design explorations, and AI predictions from our own CEO. Then we ran and a series of cross-functional ideation sessions bringing together clinical leaders, engineers, our CTO, AI consultants, and behavioral researchers. This mix ensured a vision grounded in clinical reality, technical feasibility, and human behavior.
After reviewing and theming all the ideas that had emerged, I proposed short, written narratives instead of visual mockups. Although the Head of Product was skeptical, I believed stories would better allow us to describe UI-less experiences, like background automation and voice interaction, and also help stakeholders consider key concepts and ideas without getting hung up on the details of a screen design. I used Claude to quickly synthesize the inputs into story form and called them AI vision vignettes.
AI vision vignette: AI-supported prep on the go
Maya, a Cityblock nurse, leaves her apartment at 8:40am. As she reaches the sidewalk, Cityblock’s AI assistant begins speaking through her earbuds: "Good morning! You're headed to Mrs. Ortiz first. She was discharged from the ER two days ago. I'll brief you on the way."
On the subway, the AI assistant shares the visit brief while Maya listens. She asks one question out loud, "Anything to know about the home?" and the agent responds: "Mrs. Ortiz prefers Spanish. Her daughter Sandra is her primary caretaker and is usually at home. She has a small dog, Biscuit who barks, but is friendly."
By the time she arrives, Maya’s fully prepared without ever looking at her phone.
AI vision vignette: proactive member support
Mrs. Lee picks up her device to see her weekly Cityblock AI check-in: “Good morning! You mentioned last week that mornings have been harder. How are you feeling today?” Mrs. Lee replies back: “My knees are bad today.”
The AI agent responds: “I hear you. Want me to change your afternoon appointment with your nurse, Maya, to a phone call instead of a home visit so you don’t have to be up and ready? I also noticed it’s the third time your knees have flared up. I’ll let Maya know.”
When Maya calls at 2pm, she already knows about the bad knees and is ready to followup.
The final vignettes spanned a variety of scenarios and roles, covering care team, manager and patient experiences, and considered when AI got it right and when AI got it wrong.
Despite her hesitation, our Head of Product found the stories compelling and shared them immediately with other leaders, a strong signal that the written format worked: making the vision easy to digest, discuss, and champion. Ahead of a board meeting, we selected one narrative of an AI-enhanced visit-prep to turn into a short film.
Ideas from the vision, are now entering product team H2 roadmaps.
Vision vignettes spanned a variety of scenarios and roles and considered the impact when AI got it right and when AI got it wrong.
Bringing the vision to life