Curious about what’s possible.
Grounded in problems worth solving.
I’ve always enjoyed working on complicated problems, especially where technology, healthcare, and people come together.
​
Much of my product journey has focused on digital health and virtual care.
More recently, I’ve been exploring how AI can help us think differently, make better-informed decisions, simplify complex workflows, and create new product experiences.
​
What excites me most isn’t AI for the sake of AI. It’s figuring out where it can genuinely make someone’s work or experience better, while keeping people, good judgment, and the problem we’re solving at the center.
See what I’ve been building
Featured AI Experiment
Product Visual Architect
Turning complex product information into collaboration-ready visual models
I asked a simple product question:
Could AI make it easier to turn messy product information into useful visuals without sacrificing product judgment?
I built Product Visual Architect, a Microsoft Copilot agent designed to transform product concepts, notes, requirements, customer journeys, and brainstorming into visual models while identifying gaps and assumptions. The design supports workflow, sequence, architecture, opportunity-solution, and other visual structures, with outputs intended for continued collaboration.
How it works
Product Input
↓
AI-Assisted Visual Draft
↓
Human Validation
↓
Editable Artifact
↓
Team Collaboration
I have also explored using AI-generated workflow structures as inputs for further collaboration in Visio.
The goal isn’t to let AI create the final answer. It’s to give people a better starting point for the conversation.
01 | Input
Goal: Develop a clear, step-by-step workflow for scheduled virtual visits. Distinguish the patient journey from the clinical and operational workflow, show key handoffs, and surface gaps, unknowns, or assumptions requiring stakeholder input.
Context: The workflow will support a broader virtual-care initiative and serve as the foundation for a formal diagram. The draft will be reviewed with stakeholders, refined based on feedback, and used to build alignment around the intended workflow.
Source: Synthesize relevant information from prior meeting discussions and transcripts about scheduled virtual visits. Preserve the source meaning rather than filling gaps with assumptions.
Instruction: Where information is incomplete or conflicting, flag it for validation rather than inventing an answer. Ask a clarifying question only when the missing information materially prevents a useful draft.
02 | AI Draft
03 | Human Review
Turning an AI-generated draft into something I could trust
I reviewed the first-pass workflow against the source material through four lenses:
-
Source Fidelity
-
Validated whether the workflow preserved stakeholder input, sequence, and relationships rather than assuming the AI-generated structure was correct.
-
-
Handoffs & Exceptions
-
​Reviewed whether important handoffs, exception paths, and connections between patient and operational journeys were clearly represented.
-
-
Unknowns
-
Kept unresolved questions visible for validation rather than allowing assumptions to become workflow facts.
-
-
Usability
-
Refined labels and structure so stakeholders could quickly understand, challenge, and collaborate on the workflow.
-
CLARIFIED
​
Separated patient actions from clinical and operational actions.
EXPOSED
​
Kept unresolved workflow behavior visible as
"To Validate."​
REFINED
​
Improved sequencing, handoffs, and labels to make the workflow easier to review with stakeholders.
The AI gave me a starting structure. Human review determined what was accurate, uncertain, and ready for discussion.
04 | Final Artifact
From first-pass structure to collaboration-ready workflow
[Placeholder for final diagram]
The final artifact was designed to accelerate stakeholder alignment, not replace stakeholder validation.
​
Portfolio recreation using sanitized, non-confidential content.
AI accelerates the first draft. Product judgment determines whether it is trustworthy and useful.
How I Think About Responsible AI
🔍 Preserve the Source
Don't allow assumptions to silently become facts.
âť“ Make Uncertainty Visible
Surface unknowns and areas requiring validation.
👤 Keep Humans in the Loop
AI can assist reasoning, but people remain accountable for important decisions.
🎯 Measure Usefulness
The question isn't “Can AI create this?”
It's “Did this actually help someone?”
From Building to Sharing
During a quarterly product leadership conversation, I shared several of the Copilot agents I had been building. Product Visual Architect generated particular interest, which led me to walk through my design approach and share the agent instructions as a starting point for further experimentation.
Using AI → Applying AI → Building with AI → Sharing What I’m Learning
What I’m Exploring Next
AI Products & Agents
Moving from one-off prompting toward purpose-built experiences around real user problems.
Human + AI Workflows
Understanding where AI adds value, where human judgment matters, and how the two should work together.
AI Product Evaluation
Measuring fidelity, human correction, workflow efficiency, adoption, and downstream usefulness.
What I’m Learning
The more I work with AI, the less interested I am in what it can generate on its own.
I’m much more interested in what happens when AI and good product judgment work together.
For me, that means starting with a real problem, understanding why AI belongs there, recognizing where it can fail, keeping people involved where judgment matters, and measuring whether the resulting experience actually became better.
I’m still learning, testing, building, and asking a lot of questions along the way.
Honestly, that’s what makes this part of my product journey so exciting.