Rupa Chaturvedi

Head of Merchant & Ads Design at Uber, Adjunct Lecturer at Stanford University

I’m a UX Design Leader and AI/GenAI specialist who blends 0→1 invention with platform-scale delivery. I build high-performing global teams, architect systems and information frameworks, and turn ambiguous problems into durable products that move KPIs.Highlights:- 0→1 to v2 at scale: From concept to launch to systematized patterns—GenAI assistants, video discovery, conversational shopping, AR/3D visualization.- Platform UX and IA: Unifying fragmented surfaces into responsive, multi‑platform systems; shaping primitives for workflows, analytics, and monetization.- Design org building: Scaling diverse, distributed teams; codifying operating cadences; partnering with research and content; aligning roadmaps to outcomes.- AI design enablement: Equipping UX orgs to design AI responsibly—expanding influence beyond UI via model evaluation, systems thinking, AI‑first mental models, and AI workflows.Selected outcomes:- Built a 14-person global design org across 3 continents; shipped 80+ features in under 5 months; consolidated a multi‑platform system; meaningful KPI impact.- Incubated 0→1 GenAI initiatives (consumer LLM assistant, Merchant Copilot, AI creative tools, video discovery)—adopted by SMBs and enterprises globally.- Scaled a conversational shopping product from 0→1 to 1M WAUs in under 6 months; 20+ patents granted/published; launched AR “View in Your Room/AR View” with press coverage.- Taught and enabled 1,500+ professionals via a human-centered AI institute; consistent 5/5 ratings; partners include Reforge, Stanford, Maven.Today, I focus on platform UX, AI-assisted experiences, and design org development—bringing product strategy, systems thinking, and responsible AI to enterprise and SMB workflows.

UX Design Manager

Over 4000 people signed up for my talk on Designing for Trust in AI Agents yesterday.

As the industry continues to dig deeper on what AI agents can do, the question of whether users will actually trust them to do it seems to be top of mind for many of us product builders.

I've been experiencing this firsthand. I've tested Claude Cowork, Cora, Manus, and a handful of other agents over the past few months. They're fast. They're thorough. And I keep finding myself unable to use what they produce.

The pattern I keep landing on: we've designed agents to optimize for speed to output instead of speed to understanding. The industry celebrates execution time. Users care about whether they can actually do something with the result. Those metrics have almost nothing to do with each other.

I wrote up the core ideas from the session in this week's newsletter. It goes deeper on where the trust gap actually lives and why the usual mechanisms (transparency, guardrails, reversibility) don't solve the underlying problem.

If you'd like to watch a replay of the talk, comment AIUX and I will send you the recording link.

Design & AI Leader | Stanford Lecturer | AI enablement for UX orgs | ex-Uber, Google, Amazon

This is the AI UX Brief - my weekly newsletter on designing intelligent systems, leadership in the AI era, and navigating the messy reality of AIUX. F

Open article: The Speed Paradox: Why AI Agents Feel Impressive But Unusable by Rupa Chaturvedi • 7 min read

The Speed Paradox: Why AI Agents Feel Impressive But Unusable

taplio