Designing the surface
where humans meet
AI agents.
I lead a product design team at ServiceNow working at the intersection of enterprise software, governance, and AI agents. I think of design less as the appearance of products and more as the contract between the people who use them and the systems that increasingly act on their behalf.
- Manager, Product Design at ServiceNow
- GRC · TPRM · Enterprise Architecture
- Indie practice — faith & community apps
Six projects across enterprise risk, AI-assisted workflows, design systems for AI agents, voice interfaces, and personal practice. Each one is a working argument about where this discipline is going.
- 01
An AI-native prototype for enterprise risk
A working concept for what risk and compliance look like when the manual work is automated, the data is current, and the decisions are faster.
AI-nativeExecutive productGRC2026 → - 02
Making AI-edited assessments auditable
When an AI rewrites a compliance question, who is accountable? A question-level audit trail with a floating modal, designed around how auditors actually ask their questions.
AI-nativeComplianceAuditability2025–2026 → - 03
An AI-native design system as a Claude skill
Packaging a production design system into a compact instruction file that AI coding agents can actually follow — so prototypes stop coming out almost-but-not-quite ServiceNow.
Design systemsAI toolingDeveloper experience2025–2026 → - 04
Vendor onboarding when the AI does the first draft
An AI-assisted Smart Start workflow for third-party risk — built around the insight that the assessment template, not the product around it, is what risk teams actually need help with.
AI-nativeOnboardingWorkflow2025 → - 05
Rupi, and the case for designing from a blank file
An indie family expense tracker, built solo end-to-end. The kind of work I do to stay sharp on the parts of the craft that production design systems quietly take away from you.
IndieMobilePersonal practice2024–ongoing → - 06
Voice as the first non-screen surface
A multi-device voice interface for enterprise CRM, designed by going to where the work actually happens — in cars, between meetings, anywhere except the desk the existing software was built for.
Voice UIEnterpriseCRM2018–2020 →
Most of what is interesting about
AI-native design is not the agent.
It is what has to be true around it.
Provenance is a design problem
When AI drafts a control, edits a policy, or rewrites an assessment, the trail of who-did-what becomes the most valuable artifact in the system. Designing that trail — at the right level of granularity, with the right gestures of trust — is most of the work.
Design systems are now contracts for AI agents
A design system is no longer a website for humans. It is a contract that AI agents follow when generating UI. That changes what belongs in it, how it is written, and which rules earn their place — usually fewer than you would expect.