Trustworthy AI
AI experience is state design
The prompt field is rarely the hard part. Draft, edited, verified, saved, restricted, and failed states determine whether people can understand and trust the result.
Read the AI Agents case study
Case StudiesBengaluru · India
Product design notes
Notes from enterprise product work.
Three practical observations from designing AI, data, reporting, and infrastructure products. Each note links to the case study behind it.
Enterprise UX
Governance should feel like guidance
Permissions, compatibility, and inheritance rules do not need to dominate the interface. The best systems reveal a rule at the moment it helps someone make a safe choice.
See the Global Filters systemSystems thinking
Design for the real maximum
A tidy example can make almost any enterprise pattern look successful. Testing with realistic source, field, role, and dependency counts exposes the decisions that matter.
Explore the Unified Data modelWritings on Medium
Articles & Publications
Your Colour Palette Is Not a Design System
Why basic AI color generators fail in high-density SaaS interfaces and how proper semantic token architecture scales across enterprise platforms.
I Built a Salary-First Planner for Uneven Expenses. Here’s What I Need to Validate
Designing a salary-first reserve model that smooths 12-month cash flow and eliminates surprise annual expense shortfalls.
AI Can Generate Good UI. Why Does the UX Still Feel Wrong?
Why pixel-perfect AI generated mockups crumble under real user interactions, state transitions, and edge-case validation.