Founding design leader for a 0-to-1 AI consumer recommendation marketplace driving $1B+ annual purchase volume. Built the design function from zero and owned the interaction model for the assistant end to end.
Sagely was a 0-to-1 product inside Roundforest, betting on an AI shopping assistant for an audience that had never seen the pattern. ~40M monthly users, over a billion dollars in annual purchase volume flowing through recommendations people had to trust enough to act on.
When I joined, there was no design function, no design system, no critique culture. Engineers and a product team had shipped a v1; the question now was whether design could turn it into something that felt warm and credible to a non-technical consumer.
Head of Design and founding design leader. I owned hiring, rituals, standards, design process, and the design system. I led a small in-house team and owned the core interaction model for the AI shopping experience end to end.
I partnered tightly with the data science team and the product lead. I didn't own engineering or commerce backend, but I was in the room for the calls that crossed those lines.
Trust as the central craft. For an audience that had never used an AI shopping assistant, the design problem wasn't "make the AI feel smart." It was "make the user feel safe enough to act."
We chose three primitives to design around:
The original product instinct was a chat interface. I argued against it. Chat puts the burden of asking the right question on the user, and our users didn't know what they wanted. We shifted the model to a guided flow: the AI makes the first suggestion, the user reacts, the AI adjusts. Closer to talking to a stylist than to a search bar.
We had real-time behavioral data flowing into recommendations from data science. I worked with the DS team to surface the why of each recommendation in plain language — not a confidence score, not a percentage, a sentence the user could decide whether or not to trust. That single move was the largest unlock for engagement.
I hired three designers in six months and built a critique cadence that took rounds from a week to two days. Hiring at startup pace while raising the craft bar at the same time was the operating challenge of the role — harder than any single screen.
Sagely reached ~40M monthly users and $1B+ annual purchase volume through recommendation-driven flows. The interaction model I designed for the AI assistant became the spine of the consumer experience.
The design function went from zero to a working team with shipped standards and a working critique cadence in 16 months.
The same scored recommendation, two views. Suggestion is the default, one pick at a time with a plain-language why. Grid is for shoppers who want to compare. Try the toggle.
No search box. The AI opens with its top pick. React, or scroll, and it moves to the next.
Our top pick overall. Powerful enough for ice and frozen fruit, and reviewers say it holds up for years.
Based on Amazon user reviews, YouTube videos,
expert recommendations and other reputable online resources
A powerful and versatile personal blender with stainless steel components, suitable for various blending tasks; however, its small size might not be ideal for larger batches
A powerful and versatile personal blender with stainless steel components, suitable for various blending tasks; however, its small size might not be ideal for larger batches
A powerful and versatile personal blender with stainless steel components, suitable for various blending tasks; however, its small size might not be ideal for larger batches
A powerful and versatile personal blender with stainless steel components, suitable for various blending tasks; however, its small size might not be ideal for larger batches
The recommendation surface, two views · Suggestion (default) and Grid are the same scored answer at two densities. Interactive recreation — toggle and scroll the picks.
These rankings are based on millions of shoppers' reviews, expert opinions, and online discussions across multiple platforms, including Amazon, YouTube, Google, Reddit, and more.
Best for Versatile Personal Blending. A powerful and versatile personal blender with stainless steel components, suitable for various blending tasks; however, its small size might not be ideal for larger
Drill into a pick · the product page deep-dive, with the bottom line and the per-attribute scores. Recreation for reference.
I underestimated how much research we needed before product decisions in the first quarter. I leaned on data science signals and my own intuition for too long.
Two months in, we ran a small generative study and it reshaped two assumptions I'd been holding without realizing. I'd run that study in week three, not month three.