Client: VC-backed Stealth Startup (Stanford-led)

Role: UI/UX Designer (Contract)

Timeline: April – May 2025

Platforms: Web · Mobile

I helped a stealth VC-backed startup design AI-powered shopping agents that feel more like a stylish friend than a chatbot. We focused on fashion and personal care products, delivering intuitive, human-like conversations through polished UI, smart nudges, and crafted tone.

Build a human-like AI shopping experience that feels less like a chatbot and more like a helpful, stylish friend. The startup was exploring AI agents for fashion and skincare, with the goal of making discovery and product recommendations both personal and frictionless.

Design clean, responsive UI for AI-powered shopping agents across web and mobile.

Ensure that agents are strategically placed on the screen so they support, not block, the user’s shopping experience.

Create loading, recommendation, and fallback states that feel smooth and human.

Design a flexible chat experience that supports both text and video-based conversations.

Help users understand and engage with the AI agent, and ensure the agent doesn’t feel like a chatbot, but like a real digital shopping assistant.

Brought in as the UI/UX designer during a product sprint, I worked hands-on with the founders to prototype core AI flows and craft the full user experience, from logic mapping to microcopy.

I designed the full conversation experience, including text, video, and fallback interactions.

One major challenge was ensuring the AI agent didn’t interrupt the user’s shopping flow.

I worked with the team to strategically position the agent so it stayed visible but never blocked key content, especially product cards or navigation.

This was critical for pitching to e-commerce founders in fashion and cosmetics who wanted AI help that felt personal, not pushy.

Prototyped two distinct layouts:

This gave the team flexibility to test which structure users trusted more.

2. Figma Prototypes that “Spoke AI”

Web + mobile prototypes with realistic agent responses

Enabled faster dev handoff and aligned teams on flow logic

3. Smart Nudges & UX Writing

Created inline prompts that increased engagement by 35%

Tailored tone: affirming for skincare, playful for fashion

Designed the loading transitions when users clicked the agent:

Created fallback text if the AI couldn’t respond, so the experience never felt broken.

4. Rapid Iterations with Founders

Weekly async sprints with founder feedback

Evolved the agent’s voice + UI based on user test loops

Designed the recommendation cards that appear post-chat.

Made sure users could click into one recommendation, and still have access to the rest, solving for edge cases.

Built this to work across both mobile and desktop.

2× faster dev implementation

+35% engagement with product recs

Higher user satisfaction across multiple test cycles

Conversational design is 90% tone and timing. I learned that where you place an AI agent can make or break trust, users want help, but only if it doesn't get in the way.

Let’s talk → I love projects that blend smart tech with real, helpful user flows.